graphql-engine/.circleci/test-server.sh

1446 lines
46 KiB
Bash
Raw Normal View History

#!/usr/bin/env bash
set -euo pipefail
echo "Running tests on node $CIRCLE_NODE_INDEX of $CIRCLE_NODE_TOTAL"
if [ -z "$SERVER_TEST_TO_RUN" ]; then
# shellcheck disable=SC2016
echo 'Please specify $SERVER_TEST_TO_RUN'
exit 1
else
echo "Running test $SERVER_TEST_TO_RUN"
fi
### Functions
# Disable the following warning:
# > Note that A && B || C is not if-then-else. C may run when A is true.
# We want this behavior, as we want to continue even if `kill` fails.
# shellcheck disable=SC2015
stop_services() {
echo "killing and waiting for spawned services"
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
[[ -n "${HGE_PIDS[*]}" ]] && {
kill -s INT "${HGE_PIDS[@]}"
wait "${HGE_PIDS[@]}"
} || true
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
}
time_elapsed() {
printf "(%02d:%02d)" $((SECONDS / 60)) $((SECONDS % 60))
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
}
fail_if_port_busy() {
local PORT="$1"
if nc -z localhost "$PORT"; then
echo "ERROR:"
echo "Port $PORT is busy."
echo "Output of \`lsof -i :$PORT\`:"
lsof -i ":$PORT"
echo
echo 'Exiting immediately.'
exit 1
fi
}
wait_for_port() {
local PORT="$1"
echo "waiting for $PORT"
for _ in $(seq 1 60); do
nc -z localhost "$PORT" && echo "port $PORT is ready" && return
echo -n .
sleep 0.25
done
echo "Failed waiting for $PORT" && exit 1
}
init_jwt() {
CUR_DIR="$PWD"
mkdir -p "$OUTPUT_FOLDER/ssl"
cd "$OUTPUT_FOLDER/ssl"
openssl genrsa -out jwt_private.key 2048
openssl rsa -pubout -in jwt_private.key -out jwt_public.key
openssl genpkey -algorithm ed25519 -outform PEM -out ed25519_jwt_private.key
openssl pkey -pubout -in ed25519_jwt_private.key -out ed25519_jwt_public.key
cd "$CUR_DIR"
}
# init_hge_and_test_jwt function will run the hge server using the environment varibles and run the pytest which is sent as argument
# The first argument is the relative path of the jwt-key-file. the jwt-key-file can be RSA or EdDSA
# The second argument is the test to run, eg. test_jwt_claims_map.py::TestJWTClaimsMapBasic, test_jwt.py, etc.
init_hge_and_test_jwt() {
local key_file
key_file="${OUTPUT_FOLDER}/${1}"
shift
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--hge-jwt-key-file="$key_file" --hge-jwt-conf="$HASURA_GRAPHQL_JWT_SECRET" \
"$@"
kill_hge_servers
}
init_ssl() {
mkdir -p "$OUTPUT_FOLDER/ssl"
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
(
cd "$OUTPUT_FOLDER/ssl"
openssl genrsa -out ca-key.pem 2048
openssl req -x509 -new -nodes -key ca-key.pem -days 10 -out ca.pem -subj "/CN=webhook-ca"
cp ca.pem /usr/local/share/ca-certificates/hge-tests-ca.crt
update-ca-certificates
)
}
webhook_tests_check_root() {
if [ $EUID != 0 ]; then
echo -e "webhook tests require root (in order to trust certificate authority)."
exit 1
fi
}
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
kill_hge_servers() {
kill -s INT "${HGE_PIDS[@]}" || true
wait "${HGE_PIDS[@]}" || true
HGE_PIDS=()
}
HGE_INDEX=1
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
run_hge_with_args() {
i=$((HGE_INDEX++))
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
set -x
"$GRAPHQL_ENGINE" "$@" >"$OUTPUT_FOLDER/graphql-engine-${i}.log" 2>&1 &
HGE_PIDS=("${HGE_PIDS[@]}" $!)
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
set +x
}
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
start_multiple_hge_servers() {
run_hge_with_args --database-url "$HASURA_GRAPHQL_DATABASE_URL" serve
if [ -n "${HASURA_GRAPHQL_DATABASE_URL_2:-}" ]; then
run_hge_with_args --database-url "$HASURA_GRAPHQL_DATABASE_URL_2" serve --server-port 8081
wait_for_port 8081
fi
wait_for_port 8080
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
}
source_data_sources_utils() {
# Only source this file in the $SERVER_TEST_TO_RUN case branch it's used,
# to avoid sourcing for every server job & test.
# https://github.com/hasura/graphql-engine-mono/pull/1526#discussion_r661411538
SCRIPTS_SOURCE=$CIRCLECI_FOLDER/../scripts
# shellcheck source=../scripts/data-sources-util.sh
source "$SCRIPTS_SOURCE/data-sources-util.sh"
}
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
if [ -z "${HASURA_GRAPHQL_DATABASE_URL:-}" ]; then
echo "Env var HASURA_GRAPHQL_DATABASE_URL is not set"
exit 1
fi
if [ -z "${HASURA_GRAPHQL_DATABASE_URL_2:-}" ]; then
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
echo "Env var HASURA_GRAPHQL_DATABASE_URL_2 is not set"
exit 1
fi
CIRCLECI_FOLDER="${BASH_SOURCE[0]%/*}"
cd "$CIRCLECI_FOLDER"
CIRCLECI_FOLDER="$PWD"
PYTEST_ROOT="$CIRCLECI_FOLDER/../server/tests-py"
OUTPUT_FOLDER=${OUTPUT_FOLDER:-"$CIRCLECI_FOLDER/test-server-output"}
mkdir -p "$OUTPUT_FOLDER"
cd "$PYTEST_ROOT"
for port in 8080 8081 9876 5592 5000 5001 5593 5594; do
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
fail_if_port_busy $port
done
echo -e "\nINFO: GraphQL Executable : $GRAPHQL_ENGINE"
echo -e "INFO: Logs Folder : $OUTPUT_FOLDER\n"
# Copy the node_modules directory installed in the Docker image here.
cp -R /deps/node_modules .
export EVENT_WEBHOOK_HEADER="MyEnvValue"
export HASURA_GRAPHQL_STRINGIFY_NUMERIC_TYPES=true
export DEFAULT_HASURA_EXPERIMENTAL_FEATURES=streaming_subscriptions
export HASURA_GRAPHQL_EXPERIMENTAL_FEATURES=$DEFAULT_HASURA_EXPERIMENTAL_FEATURES
export HGE_URL="http://localhost:8080"
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
export HGE_URL_2=""
if [ -n "${HASURA_GRAPHQL_DATABASE_URL_2:-}" ]; then
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
HGE_URL_2="http://localhost:8081"
fi
export EVENT_WEBHOOK_HEADER="MyEnvValue"
export EVENT_WEBHOOK_HANDLER="http://localhost:5592"
export ACTION_WEBHOOK_HANDLER="http://localhost:5593"
export SCHEDULED_TRIGGERS_WEBHOOK_DOMAIN="http://localhost:5594"
export REMOTE_SCHEMAS_WEBHOOK_DOMAIN="http://localhost:5000"
export GRAPHQL_SERVICE_HANDLER="http://localhost:4001"
export GRAPHQL_SERVICE_1="http://localhost:4020"
export GRAPHQL_SERVICE_2="http://localhost:4021"
export GRAPHQL_SERVICE_3="http://localhost:4022"
PYTEST_REPORTING_ARGS=(
'-vv' # show all passes and the full assertion text
'-r' 'a' # show extra test summary info for every test except passes
)
PYTEST_COMMON_ARGS=(
"${PYTEST_REPORTING_ARGS[@]}"
--hge-urls="$HGE_URL"
--pg-urls="$HASURA_GRAPHQL_DATABASE_URL"
)
PYTEST_PARALLEL_ARGS=(
"${PYTEST_REPORTING_ARGS[@]}"
-n 2
--hge-urls "$HGE_URL" "${HGE_URL_2:-}"
--pg-urls "$HASURA_GRAPHQL_DATABASE_URL" "${HASURA_GRAPHQL_DATABASE_URL_2}"
)
HGE_PIDS=()
trap stop_services ERR
trap stop_services INT
trap stop_services EXIT
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
run_pytest_parallel() {
trap stop_services ERR
if [ -n "${HASURA_GRAPHQL_DATABASE_URL_2:-}" ]; then
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
set -x
pytest "${PYTEST_PARALLEL_ARGS[@]}" "$@"
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
set +x
else
set -x
pytest "${PYTEST_COMMON_ARGS[@]}" "$@"
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
set +x
fi
}
case "$SERVER_TEST_TO_RUN" in
no-auth)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITHOUT ADMIN SECRET ###########################################>\n"
start_multiple_hge_servers
run_pytest_parallel
kill_hge_servers
;;
admin-secret)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET #####################################>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
start_multiple_hge_servers
run_pytest_parallel
kill_hge_servers
;;
admin-secret-unauthorized-role)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND UNAUTHORIZED ROLE #####################################>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_UNAUTHORIZED_ROLE="anonymous"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-unauthorized-role \
test_graphql_queries.py::TestUnauthorizedRolePermission
kill_hge_servers
#unset HASURA_GRAPHQL_UNAUTHORIZED_ROLE
;;
jwt-rs512)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (RS512) #####################################>\n"
init_jwt
HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key }')"
export HASURA_GRAPHQL_ADMIN_SECRET HASURA_GRAPHQL_JWT_SECRET
start_multiple_hge_servers
run_pytest_parallel --hge-jwt-key-file="$OUTPUT_FOLDER/ssl/jwt_private.key" --hge-jwt-conf="$HASURA_GRAPHQL_JWT_SECRET"
kill_hge_servers
#unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-ed25519)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (Ed25519) #####################################>\n"
init_jwt
HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key }')"
export HASURA_GRAPHQL_ADMIN_SECRET HASURA_GRAPHQL_JWT_SECRET
start_multiple_hge_servers
run_pytest_parallel --hge-jwt-key-file="$OUTPUT_FOLDER/ssl/ed25519_jwt_private.key" --hge-jwt-conf="$HASURA_GRAPHQL_JWT_SECRET"
kill_hge_servers
#unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-stringified)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (in stringified mode) #####################################>\n"
init_jwt
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , claims_format: "stringified_json"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt.py
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , claims_format: "stringified_json"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt.py
# unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-audience-check-single-string)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (with audience check - string) #####################################>\n"
init_jwt
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , audience: "myapp-1234"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt.py
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , audience: "myapp-1234"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt.py
#unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-audience-check-list-string)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (with audience check - list of strings) #################################>\n"
init_jwt
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , audience: ["myapp-1234", "myapp-9876"]}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt.py
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , audience: ["myapp-1234", "myapp-9876"]}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt.py
unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-issuer-check)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (with issuer check) #####################################>\n"
init_jwt
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , issuer: "https://hasura.com"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt.py
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , issuer: "https://hasura.com"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt.py
unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-with-claims-namespace-path)
##########
# TODO(swann): should these not be run in parallel?
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (with claims_namespace_path) #####################################>\n"
init_jwt
# hasura claims at one level of nesting
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , claims_namespace_path: "$.hasura_claims"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt.py
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , claims_namespace_path: "$.hasura_claims"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt.py
unset HASURA_GRAPHQL_JWT_SECRET
# hasura claims at two levels of nesting with claims_namespace_path containing special character
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , claims_namespace_path: "$.hasura['\''claims%'\'']"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt.py
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , claims_namespace_path: "$.hasura['\''claims%'\'']"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt.py
unset HASURA_GRAPHQL_JWT_SECRET
# hasura claims at the root of the JWT token
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , claims_namespace_path: "$"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt.py
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , claims_namespace_path: "$"}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt.py
unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-claims-map-with-json-path-values)
# test JWT with Claims map
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (with claims_map and values are json path) #####################################>\n"
init_jwt
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , claims_map: {"x-hasura-user-id": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].user.id"}, "x-hasura-allowed-roles": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].role.allowed"}, "x-hasura-default-role": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].role.default"}}}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt_claims_map.py::TestJWTClaimsMapBasic
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , claims_map: {"x-hasura-user-id": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].user.id"}, "x-hasura-allowed-roles": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].role.allowed"}, "x-hasura-default-role": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].role.default"}}}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt_claims_map.py::TestJWTClaimsMapBasic
unset HASURA_GRAPHQL_JWT_SECRET
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (with claims_map and values are json path with default values set) #####################################>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , claims_map: {"x-hasura-user-id": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].user.id", "default":"1"}, "x-hasura-allowed-roles": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].role.allowed", "default":["user","editor"]}, "x-hasura-default-role": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].role.default","default":"user"}}}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt_claims_map.py::TestJWTClaimsMapBasic
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , claims_map: {"x-hasura-user-id": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].user.id", "default":"1"}, "x-hasura-allowed-roles": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].role.allowed", "default":["user","editor"]}, "x-hasura-default-role": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].role.default","default":"user"}}}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt_claims_map.py::TestJWTClaimsMapBasic
unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-with-expiry-time-leeway)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (with JWT config allowing for leeway) #####################################>\n"
init_jwt
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , allowed_skew: 60}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt.py::TestJWTExpirySkew
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , allowed_skew: 60}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt.py::TestJWTExpirySkew
unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-claims-map-with-literal-values)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (with claims_map and values are literal values) #####################################>\n"
init_jwt
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , claims_map: {"x-hasura-user-id": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].user.id"}, "x-hasura-allowed-roles": ["user","editor"], "x-hasura-default-role": "user","x-hasura-custom-header":"custom-value"}}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt_claims_map.py::TestJWTClaimsMapWithStaticHasuraClaimsMapValues
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , claims_map: {"x-hasura-user-id": {"path":"$.['"'"'https://myapp.com/jwt/claims'"'"'].user.id"}, "x-hasura-allowed-roles": ["user","editor"], "x-hasura-default-role": "user","x-hasura-custom-header":"custom-value"}}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt_claims_map.py::TestJWTClaimsMapWithStaticHasuraClaimsMapValues
unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-cookie)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET AND JWT (in cookie mode) #####################################>\n"
init_jwt
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , header: {"type": "Cookie", "name": "hasura_user"}}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/jwt_private.key" test_jwt.py
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/ed25519_jwt_public.key")" '{ type: "Ed25519", key: $key , header: {"type": "Cookie", "name": "hasura_user"}}')"
export HASURA_GRAPHQL_JWT_SECRET
init_hge_and_test_jwt "ssl/ed25519_jwt_private.key" test_jwt.py
unset HASURA_GRAPHQL_JWT_SECRET
;;
jwt-cookie-unauthorized-role)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH JWT (in cookie mode) AND UNAUTHORIZED ROLE #####################################>\n"
init_jwt
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_UNAUTHORIZED_ROLE="anonymous"
HASURA_GRAPHQL_JWT_SECRET="$(jq -n --arg key "$(cat "$OUTPUT_FOLDER/ssl/jwt_public.key")" '{ type: "RS512", key: $key , header: {"type": "Cookie", "name": "hasura_user"}}')"
export HASURA_GRAPHQL_JWT_SECRET
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--hge-jwt-key-file="$OUTPUT_FOLDER/ssl/jwt_private.key" --hge-jwt-conf="$HASURA_GRAPHQL_JWT_SECRET" \
--test-unauthorized-role \
test_graphql_queries.py::TestFallbackUnauthorizedRoleCookie
kill_hge_servers
unset HASURA_GRAPHQL_UNAUTHORIZED_ROLE
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--hge-jwt-key-file="$OUTPUT_FOLDER/ssl/jwt_private.key" --hge-jwt-conf="$HASURA_GRAPHQL_JWT_SECRET" \
--test-no-cookie-and-unauth-role \
test_graphql_queries.py::TestMissingUnauthorizedRoleAndCookie
kill_hge_servers
server: forward auth webhook set-cookies header on response > High-Level TODO: * [x] Code Changes * [x] Tests * [x] Check that pro/multitenant build ok * [x] Documentation Changes * [x] Updating this PR with full details * [ ] Reviews * [ ] Ensure code has all FIXMEs and TODOs addressed * [x] Ensure no files are checked in mistakenly * [x] Consider impact on console, cli, etc. ### Description > This PR adds support for adding set-cookie header on the response from the auth webhook. If the set-cookie header is sent by the webhook, it will be forwarded in the graphQL engine response. Fixes a bug in test-server.sh: testing of get-webhook tests was done by POST method and vice versa. To fix, the parameters were swapped. ### Changelog - [x] `CHANGELOG.md` is updated with user-facing content relevant to this PR. ### Affected components - [x] Server - [ ] Console - [ ] CLI - [x] Docs - [ ] Community Content - [ ] Build System - [x] Tests - [ ] Other (list it) ### Related Issues -> Closes [#2269](https://github.com/hasura/graphql-engine/issues/2269) ### Solution and Design > ### Steps to test and verify > Please refer to the docs to see how to send the set-cookie header from webhook. ### Limitations, known bugs & workarounds > - Support for only set-cookie header forwarding is added - the value forwarded in the set-cookie header cannot be validated completely, the [Cookie](https://hackage.haskell.org/package/cookie) package has been used to parse the header value and any unnecessary information is stripped off before forwarding the header. The standard given in [RFC6265](https://datatracker.ietf.org/doc/html/rfc6265) has been followed for the Set-Cookie format. ### Server checklist #### Catalog upgrade Does this PR change Hasura Catalog version? - [x] No - [ ] Yes - [ ] Updated docs with SQL for downgrading the catalog #### Metadata Does this PR add a new Metadata feature? - [x] No #### GraphQL - [x] No new GraphQL schema is generated - [ ] New GraphQL schema is being generated: - [ ] New types and typenames are correlated #### Breaking changes - [x] No Breaking changes PR-URL: https://github.com/hasura/graphql-engine-mono/pull/2538 Co-authored-by: Robert <132113+robx@users.noreply.github.com> GitOrigin-RevId: d9047e997dd221b7ce4fef51911c3694037e7c3f
2021-11-09 15:00:21 +03:00
unset HASURA_GRAPHQL_JWT_SECRET
;;
server: forward auth webhook set-cookies header on response > High-Level TODO: * [x] Code Changes * [x] Tests * [x] Check that pro/multitenant build ok * [x] Documentation Changes * [x] Updating this PR with full details * [ ] Reviews * [ ] Ensure code has all FIXMEs and TODOs addressed * [x] Ensure no files are checked in mistakenly * [x] Consider impact on console, cli, etc. ### Description > This PR adds support for adding set-cookie header on the response from the auth webhook. If the set-cookie header is sent by the webhook, it will be forwarded in the graphQL engine response. Fixes a bug in test-server.sh: testing of get-webhook tests was done by POST method and vice versa. To fix, the parameters were swapped. ### Changelog - [x] `CHANGELOG.md` is updated with user-facing content relevant to this PR. ### Affected components - [x] Server - [ ] Console - [ ] CLI - [x] Docs - [ ] Community Content - [ ] Build System - [x] Tests - [ ] Other (list it) ### Related Issues -> Closes [#2269](https://github.com/hasura/graphql-engine/issues/2269) ### Solution and Design > ### Steps to test and verify > Please refer to the docs to see how to send the set-cookie header from webhook. ### Limitations, known bugs & workarounds > - Support for only set-cookie header forwarding is added - the value forwarded in the set-cookie header cannot be validated completely, the [Cookie](https://hackage.haskell.org/package/cookie) package has been used to parse the header value and any unnecessary information is stripped off before forwarding the header. The standard given in [RFC6265](https://datatracker.ietf.org/doc/html/rfc6265) has been followed for the Set-Cookie format. ### Server checklist #### Catalog upgrade Does this PR change Hasura Catalog version? - [x] No - [ ] Yes - [ ] Updated docs with SQL for downgrading the catalog #### Metadata Does this PR add a new Metadata feature? - [x] No #### GraphQL - [x] No new GraphQL schema is generated - [ ] New GraphQL schema is being generated: - [ ] New types and typenames are correlated #### Breaking changes - [x] No Breaking changes PR-URL: https://github.com/hasura/graphql-engine-mono/pull/2538 Co-authored-by: Robert <132113+robx@users.noreply.github.com> GitOrigin-RevId: d9047e997dd221b7ce4fef51911c3694037e7c3f
2021-11-09 15:00:21 +03:00
# test with CORS modes
cors-domains)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH CORS DOMAINS ########>\n"
export HASURA_GRAPHQL_CORS_DOMAIN="http://*.localhost, http://localhost:3000, https://*.foo.bar.com"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
server: forward auth webhook set-cookies header on response > High-Level TODO: * [x] Code Changes * [x] Tests * [x] Check that pro/multitenant build ok * [x] Documentation Changes * [x] Updating this PR with full details * [ ] Reviews * [ ] Ensure code has all FIXMEs and TODOs addressed * [x] Ensure no files are checked in mistakenly * [x] Consider impact on console, cli, etc. ### Description > This PR adds support for adding set-cookie header on the response from the auth webhook. If the set-cookie header is sent by the webhook, it will be forwarded in the graphQL engine response. Fixes a bug in test-server.sh: testing of get-webhook tests was done by POST method and vice versa. To fix, the parameters were swapped. ### Changelog - [x] `CHANGELOG.md` is updated with user-facing content relevant to this PR. ### Affected components - [x] Server - [ ] Console - [ ] CLI - [x] Docs - [ ] Community Content - [ ] Build System - [x] Tests - [ ] Other (list it) ### Related Issues -> Closes [#2269](https://github.com/hasura/graphql-engine/issues/2269) ### Solution and Design > ### Steps to test and verify > Please refer to the docs to see how to send the set-cookie header from webhook. ### Limitations, known bugs & workarounds > - Support for only set-cookie header forwarding is added - the value forwarded in the set-cookie header cannot be validated completely, the [Cookie](https://hackage.haskell.org/package/cookie) package has been used to parse the header value and any unnecessary information is stripped off before forwarding the header. The standard given in [RFC6265](https://datatracker.ietf.org/doc/html/rfc6265) has been followed for the Set-Cookie format. ### Server checklist #### Catalog upgrade Does this PR change Hasura Catalog version? - [x] No - [ ] Yes - [ ] Updated docs with SQL for downgrading the catalog #### Metadata Does this PR add a new Metadata feature? - [x] No #### GraphQL - [x] No new GraphQL schema is generated - [ ] New GraphQL schema is being generated: - [ ] New types and typenames are correlated #### Breaking changes - [x] No Breaking changes PR-URL: https://github.com/hasura/graphql-engine-mono/pull/2538 Co-authored-by: Robert <132113+robx@users.noreply.github.com> GitOrigin-RevId: d9047e997dd221b7ce4fef51911c3694037e7c3f
2021-11-09 15:00:21 +03:00
run_hge_with_args serve
wait_for_port 8080
server: forward auth webhook set-cookies header on response > High-Level TODO: * [x] Code Changes * [x] Tests * [x] Check that pro/multitenant build ok * [x] Documentation Changes * [x] Updating this PR with full details * [ ] Reviews * [ ] Ensure code has all FIXMEs and TODOs addressed * [x] Ensure no files are checked in mistakenly * [x] Consider impact on console, cli, etc. ### Description > This PR adds support for adding set-cookie header on the response from the auth webhook. If the set-cookie header is sent by the webhook, it will be forwarded in the graphQL engine response. Fixes a bug in test-server.sh: testing of get-webhook tests was done by POST method and vice versa. To fix, the parameters were swapped. ### Changelog - [x] `CHANGELOG.md` is updated with user-facing content relevant to this PR. ### Affected components - [x] Server - [ ] Console - [ ] CLI - [x] Docs - [ ] Community Content - [ ] Build System - [x] Tests - [ ] Other (list it) ### Related Issues -> Closes [#2269](https://github.com/hasura/graphql-engine/issues/2269) ### Solution and Design > ### Steps to test and verify > Please refer to the docs to see how to send the set-cookie header from webhook. ### Limitations, known bugs & workarounds > - Support for only set-cookie header forwarding is added - the value forwarded in the set-cookie header cannot be validated completely, the [Cookie](https://hackage.haskell.org/package/cookie) package has been used to parse the header value and any unnecessary information is stripped off before forwarding the header. The standard given in [RFC6265](https://datatracker.ietf.org/doc/html/rfc6265) has been followed for the Set-Cookie format. ### Server checklist #### Catalog upgrade Does this PR change Hasura Catalog version? - [x] No - [ ] Yes - [ ] Updated docs with SQL for downgrading the catalog #### Metadata Does this PR add a new Metadata feature? - [x] No #### GraphQL - [x] No new GraphQL schema is generated - [ ] New GraphQL schema is being generated: - [ ] New types and typenames are correlated #### Breaking changes - [x] No Breaking changes PR-URL: https://github.com/hasura/graphql-engine-mono/pull/2538 Co-authored-by: Robert <132113+robx@users.noreply.github.com> GitOrigin-RevId: d9047e997dd221b7ce4fef51911c3694037e7c3f
2021-11-09 15:00:21 +03:00
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_cors.py
server: forward auth webhook set-cookies header on response > High-Level TODO: * [x] Code Changes * [x] Tests * [x] Check that pro/multitenant build ok * [x] Documentation Changes * [x] Updating this PR with full details * [ ] Reviews * [ ] Ensure code has all FIXMEs and TODOs addressed * [x] Ensure no files are checked in mistakenly * [x] Consider impact on console, cli, etc. ### Description > This PR adds support for adding set-cookie header on the response from the auth webhook. If the set-cookie header is sent by the webhook, it will be forwarded in the graphQL engine response. Fixes a bug in test-server.sh: testing of get-webhook tests was done by POST method and vice versa. To fix, the parameters were swapped. ### Changelog - [x] `CHANGELOG.md` is updated with user-facing content relevant to this PR. ### Affected components - [x] Server - [ ] Console - [ ] CLI - [x] Docs - [ ] Community Content - [ ] Build System - [x] Tests - [ ] Other (list it) ### Related Issues -> Closes [#2269](https://github.com/hasura/graphql-engine/issues/2269) ### Solution and Design > ### Steps to test and verify > Please refer to the docs to see how to send the set-cookie header from webhook. ### Limitations, known bugs & workarounds > - Support for only set-cookie header forwarding is added - the value forwarded in the set-cookie header cannot be validated completely, the [Cookie](https://hackage.haskell.org/package/cookie) package has been used to parse the header value and any unnecessary information is stripped off before forwarding the header. The standard given in [RFC6265](https://datatracker.ietf.org/doc/html/rfc6265) has been followed for the Set-Cookie format. ### Server checklist #### Catalog upgrade Does this PR change Hasura Catalog version? - [x] No - [ ] Yes - [ ] Updated docs with SQL for downgrading the catalog #### Metadata Does this PR add a new Metadata feature? - [x] No #### GraphQL - [x] No new GraphQL schema is generated - [ ] New GraphQL schema is being generated: - [ ] New types and typenames are correlated #### Breaking changes - [x] No Breaking changes PR-URL: https://github.com/hasura/graphql-engine-mono/pull/2538 Co-authored-by: Robert <132113+robx@users.noreply.github.com> GitOrigin-RevId: d9047e997dd221b7ce4fef51911c3694037e7c3f
2021-11-09 15:00:21 +03:00
kill_hge_servers
unset HASURA_GRAPHQL_CORS_DOMAIN
;;
auth-webhook-cookie)
# test auth webhook set-cookie forwarding on response
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH SET-COOKIE HEADER IN AUTH WEBHOOK ########>\n"
export HASURA_GRAPHQL_AUTH_HOOK="http://localhost:9876/auth"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_auth_webhook_cookie.py
kill_hge_servers
;;
ws-init-cookie-read-cors-enabled)
# test websocket transport with initial cookie header
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH COOKIE IN WEBSOCKET INIT ########>\n"
export HASURA_GRAPHQL_AUTH_HOOK="http://localhost:9876/auth"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="POST"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
run_hge_with_args serve
wait_for_port 8080
echo "$(time_elapsed): testcase 1: read cookie, cors enabled"
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-ws-init-cookie=read \
test_websocket_init_cookie.py
kill_hge_servers
;;
ws-init-cookie-noread)
echo "$(time_elapsed): testcase 2: no read cookie, cors disabled"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_AUTH_HOOK="http://localhost:9876/auth"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="POST"
run_hge_with_args serve --disable-cors
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-ws-init-cookie=noread \
test_websocket_init_cookie.py
kill_hge_servers
;;
ws-init-cookie-read-cors-disabled)
echo "$(time_elapsed): testcase 3: read cookie, cors disabled and ws-read-cookie"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_AUTH_HOOK="http://localhost:9876/auth"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="POST"
export HASURA_GRAPHQL_WS_READ_COOKIE="true"
run_hge_with_args serve --disable-cors
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-ws-init-cookie=read \
test_websocket_init_cookie.py
kill_hge_servers
;;
ws-graphql-api-disabled)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH GRAPHQL DISABLED ########>\n"
export HASURA_GRAPHQL_ENABLED_APIS="metadata"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_AUTH_HOOK="http://localhost:9876/auth"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="POST"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_apis_disabled.py
kill_hge_servers
;;
ws-metadata-api-disabled)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH METADATA DISABLED ########>\n"
export HASURA_GRAPHQL_ENABLED_APIS="graphql"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_AUTH_HOOK="http://localhost:9876/auth"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="POST"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_apis_disabled.py
kill_hge_servers
;;
remote-schema-permissions)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH REMOTE SCHEMA PERMISSIONS ENABLED ########>\n"
export HASURA_GRAPHQL_ENABLE_REMOTE_SCHEMA_PERMISSIONS=true
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_remote_schema_permissions.py
unset HASURA_GRAPHQL_ENABLE_REMOTE_SCHEMA_PERMISSIONS
[Preview] Inherited roles for postgres read queries fixes #3868 docker image - `hasura/graphql-engine:inherited-roles-preview-48b73a2de` Note: To be able to use the inherited roles feature, the graphql-engine should be started with the env variable `HASURA_GRAPHQL_EXPERIMENTAL_FEATURES` set to `inherited_roles`. Introduction ------------ This PR implements the idea of multiple roles as presented in this [paper](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/FGALanguageICDE07.pdf). The multiple roles feature in this PR can be used via inherited roles. An inherited role is a role which can be created by combining multiple singular roles. For example, if there are two roles `author` and `editor` configured in the graphql-engine, then we can create a inherited role with the name of `combined_author_editor` role which will combine the select permissions of the `author` and `editor` roles and then make GraphQL queries using the `combined_author_editor`. How are select permissions of different roles are combined? ------------------------------------------------------------ A select permission includes 5 things: 1. Columns accessible to the role 2. Row selection filter 3. Limit 4. Allow aggregation 5. Scalar computed fields accessible to the role Suppose there are two roles, `role1` gives access to the `address` column with row filter `P1` and `role2` gives access to both the `address` and the `phone` column with row filter `P2` and we create a new role `combined_roles` which combines `role1` and `role2`. Let's say the following GraphQL query is queried with the `combined_roles` role. ```graphql query { employees { address phone } } ``` This will translate to the following SQL query: ```sql select (case when (P1 or P2) then address else null end) as address, (case when P2 then phone else null end) as phone from employee where (P1 or P2) ``` The other parameters of the select permission will be combined in the following manner: 1. Limit - Minimum of the limits will be the limit of the inherited role 2. Allow aggregations - If any of the role allows aggregation, then the inherited role will allow aggregation 3. Scalar computed fields - same as table column fields, as in the above example APIs for inherited roles: ---------------------- 1. `add_inherited_role` `add_inherited_role` is the [metadata API](https://hasura.io/docs/1.0/graphql/core/api-reference/index.html#schema-metadata-api) to create a new inherited role. It accepts two arguments `role_name`: the name of the inherited role to be added (String) `role_set`: list of roles that need to be combined (Array of Strings) Example: ```json { "type": "add_inherited_role", "args": { "role_name":"combined_user", "role_set":[ "user", "user1" ] } } ``` After adding the inherited role, the inherited role can be used like single roles like earlier Note: An inherited role can only be created with non-inherited/singular roles. 2. `drop_inherited_role` The `drop_inherited_role` API accepts the name of the inherited role and drops it from the metadata. It accepts a single argument: `role_name`: name of the inherited role to be dropped Example: ```json { "type": "drop_inherited_role", "args": { "role_name":"combined_user" } } ``` Metadata --------- The derived roles metadata will be included under the `experimental_features` key while exporting the metadata. ```json { "experimental_features": { "derived_roles": [ { "role_name": "manager_is_employee_too", "role_set": [ "employee", "manager" ] } ] } } ``` Scope ------ Only postgres queries and subscriptions are supported in this PR. Important points: ----------------- 1. All columns exposed to an inherited role will be marked as `nullable`, this is done so that cell value nullification can be done. TODOs ------- - [ ] Tests - [ ] Test a GraphQL query running with a inherited role without enabling inherited roles in experimental features - [] Tests for aggregate queries, limit, computed fields, functions, subscriptions (?) - [ ] Introspection test with a inherited role (nullability changes in a inherited role) - [ ] Docs - [ ] Changelog Co-authored-by: Vamshi Surabhi <6562944+0x777@users.noreply.github.com> GitOrigin-RevId: 3b8ee1e11f5ceca80fe294f8c074d42fbccfec63
2021-03-08 14:14:13 +03:00
kill_hge_servers
;;
[Preview] Inherited roles for postgres read queries fixes #3868 docker image - `hasura/graphql-engine:inherited-roles-preview-48b73a2de` Note: To be able to use the inherited roles feature, the graphql-engine should be started with the env variable `HASURA_GRAPHQL_EXPERIMENTAL_FEATURES` set to `inherited_roles`. Introduction ------------ This PR implements the idea of multiple roles as presented in this [paper](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/FGALanguageICDE07.pdf). The multiple roles feature in this PR can be used via inherited roles. An inherited role is a role which can be created by combining multiple singular roles. For example, if there are two roles `author` and `editor` configured in the graphql-engine, then we can create a inherited role with the name of `combined_author_editor` role which will combine the select permissions of the `author` and `editor` roles and then make GraphQL queries using the `combined_author_editor`. How are select permissions of different roles are combined? ------------------------------------------------------------ A select permission includes 5 things: 1. Columns accessible to the role 2. Row selection filter 3. Limit 4. Allow aggregation 5. Scalar computed fields accessible to the role Suppose there are two roles, `role1` gives access to the `address` column with row filter `P1` and `role2` gives access to both the `address` and the `phone` column with row filter `P2` and we create a new role `combined_roles` which combines `role1` and `role2`. Let's say the following GraphQL query is queried with the `combined_roles` role. ```graphql query { employees { address phone } } ``` This will translate to the following SQL query: ```sql select (case when (P1 or P2) then address else null end) as address, (case when P2 then phone else null end) as phone from employee where (P1 or P2) ``` The other parameters of the select permission will be combined in the following manner: 1. Limit - Minimum of the limits will be the limit of the inherited role 2. Allow aggregations - If any of the role allows aggregation, then the inherited role will allow aggregation 3. Scalar computed fields - same as table column fields, as in the above example APIs for inherited roles: ---------------------- 1. `add_inherited_role` `add_inherited_role` is the [metadata API](https://hasura.io/docs/1.0/graphql/core/api-reference/index.html#schema-metadata-api) to create a new inherited role. It accepts two arguments `role_name`: the name of the inherited role to be added (String) `role_set`: list of roles that need to be combined (Array of Strings) Example: ```json { "type": "add_inherited_role", "args": { "role_name":"combined_user", "role_set":[ "user", "user1" ] } } ``` After adding the inherited role, the inherited role can be used like single roles like earlier Note: An inherited role can only be created with non-inherited/singular roles. 2. `drop_inherited_role` The `drop_inherited_role` API accepts the name of the inherited role and drops it from the metadata. It accepts a single argument: `role_name`: name of the inherited role to be dropped Example: ```json { "type": "drop_inherited_role", "args": { "role_name":"combined_user" } } ``` Metadata --------- The derived roles metadata will be included under the `experimental_features` key while exporting the metadata. ```json { "experimental_features": { "derived_roles": [ { "role_name": "manager_is_employee_too", "role_set": [ "employee", "manager" ] } ] } } ``` Scope ------ Only postgres queries and subscriptions are supported in this PR. Important points: ----------------- 1. All columns exposed to an inherited role will be marked as `nullable`, this is done so that cell value nullification can be done. TODOs ------- - [ ] Tests - [ ] Test a GraphQL query running with a inherited role without enabling inherited roles in experimental features - [] Tests for aggregate queries, limit, computed fields, functions, subscriptions (?) - [ ] Introspection test with a inherited role (nullability changes in a inherited role) - [ ] Docs - [ ] Changelog Co-authored-by: Vamshi Surabhi <6562944+0x777@users.noreply.github.com> GitOrigin-RevId: 3b8ee1e11f5ceca80fe294f8c074d42fbccfec63
2021-03-08 14:14:13 +03:00
remote-schema-prioritize-data)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH REMOTE SCHEMA PRIORITIZE DATA/ERRORS ########>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_remote_schema_prioritize_none.py
kill_hge_servers
export HASURA_GRAPHQL_REMOTE_SCHEMA_PRIORITIZE_DATA=true
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_remote_schema_prioritize_data.py
unset HASURA_GRAPHQL_REMOTE_SCHEMA_PRIORITIZE_DATA
kill_hge_servers
;;
function-permissions)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH FUNCTION PERMISSIONS ENABLED ########>\n"
export HASURA_GRAPHQL_INFER_FUNCTION_PERMISSIONS=false
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
run_hge_with_args serve
wait_for_port 8080
[Preview] Inherited roles for postgres read queries fixes #3868 docker image - `hasura/graphql-engine:inherited-roles-preview-48b73a2de` Note: To be able to use the inherited roles feature, the graphql-engine should be started with the env variable `HASURA_GRAPHQL_EXPERIMENTAL_FEATURES` set to `inherited_roles`. Introduction ------------ This PR implements the idea of multiple roles as presented in this [paper](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/FGALanguageICDE07.pdf). The multiple roles feature in this PR can be used via inherited roles. An inherited role is a role which can be created by combining multiple singular roles. For example, if there are two roles `author` and `editor` configured in the graphql-engine, then we can create a inherited role with the name of `combined_author_editor` role which will combine the select permissions of the `author` and `editor` roles and then make GraphQL queries using the `combined_author_editor`. How are select permissions of different roles are combined? ------------------------------------------------------------ A select permission includes 5 things: 1. Columns accessible to the role 2. Row selection filter 3. Limit 4. Allow aggregation 5. Scalar computed fields accessible to the role Suppose there are two roles, `role1` gives access to the `address` column with row filter `P1` and `role2` gives access to both the `address` and the `phone` column with row filter `P2` and we create a new role `combined_roles` which combines `role1` and `role2`. Let's say the following GraphQL query is queried with the `combined_roles` role. ```graphql query { employees { address phone } } ``` This will translate to the following SQL query: ```sql select (case when (P1 or P2) then address else null end) as address, (case when P2 then phone else null end) as phone from employee where (P1 or P2) ``` The other parameters of the select permission will be combined in the following manner: 1. Limit - Minimum of the limits will be the limit of the inherited role 2. Allow aggregations - If any of the role allows aggregation, then the inherited role will allow aggregation 3. Scalar computed fields - same as table column fields, as in the above example APIs for inherited roles: ---------------------- 1. `add_inherited_role` `add_inherited_role` is the [metadata API](https://hasura.io/docs/1.0/graphql/core/api-reference/index.html#schema-metadata-api) to create a new inherited role. It accepts two arguments `role_name`: the name of the inherited role to be added (String) `role_set`: list of roles that need to be combined (Array of Strings) Example: ```json { "type": "add_inherited_role", "args": { "role_name":"combined_user", "role_set":[ "user", "user1" ] } } ``` After adding the inherited role, the inherited role can be used like single roles like earlier Note: An inherited role can only be created with non-inherited/singular roles. 2. `drop_inherited_role` The `drop_inherited_role` API accepts the name of the inherited role and drops it from the metadata. It accepts a single argument: `role_name`: name of the inherited role to be dropped Example: ```json { "type": "drop_inherited_role", "args": { "role_name":"combined_user" } } ``` Metadata --------- The derived roles metadata will be included under the `experimental_features` key while exporting the metadata. ```json { "experimental_features": { "derived_roles": [ { "role_name": "manager_is_employee_too", "role_set": [ "employee", "manager" ] } ] } } ``` Scope ------ Only postgres queries and subscriptions are supported in this PR. Important points: ----------------- 1. All columns exposed to an inherited role will be marked as `nullable`, this is done so that cell value nullification can be done. TODOs ------- - [ ] Tests - [ ] Test a GraphQL query running with a inherited role without enabling inherited roles in experimental features - [] Tests for aggregate queries, limit, computed fields, functions, subscriptions (?) - [ ] Introspection test with a inherited role (nullability changes in a inherited role) - [ ] Docs - [ ] Changelog Co-authored-by: Vamshi Surabhi <6562944+0x777@users.noreply.github.com> GitOrigin-RevId: 3b8ee1e11f5ceca80fe294f8c074d42fbccfec63
2021-03-08 14:14:13 +03:00
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_graphql_queries.py::TestGraphQLQueryFunctionPermissions
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_graphql_mutations.py::TestGraphQLMutationFunctions
[Preview] Inherited roles for postgres read queries fixes #3868 docker image - `hasura/graphql-engine:inherited-roles-preview-48b73a2de` Note: To be able to use the inherited roles feature, the graphql-engine should be started with the env variable `HASURA_GRAPHQL_EXPERIMENTAL_FEATURES` set to `inherited_roles`. Introduction ------------ This PR implements the idea of multiple roles as presented in this [paper](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/FGALanguageICDE07.pdf). The multiple roles feature in this PR can be used via inherited roles. An inherited role is a role which can be created by combining multiple singular roles. For example, if there are two roles `author` and `editor` configured in the graphql-engine, then we can create a inherited role with the name of `combined_author_editor` role which will combine the select permissions of the `author` and `editor` roles and then make GraphQL queries using the `combined_author_editor`. How are select permissions of different roles are combined? ------------------------------------------------------------ A select permission includes 5 things: 1. Columns accessible to the role 2. Row selection filter 3. Limit 4. Allow aggregation 5. Scalar computed fields accessible to the role Suppose there are two roles, `role1` gives access to the `address` column with row filter `P1` and `role2` gives access to both the `address` and the `phone` column with row filter `P2` and we create a new role `combined_roles` which combines `role1` and `role2`. Let's say the following GraphQL query is queried with the `combined_roles` role. ```graphql query { employees { address phone } } ``` This will translate to the following SQL query: ```sql select (case when (P1 or P2) then address else null end) as address, (case when P2 then phone else null end) as phone from employee where (P1 or P2) ``` The other parameters of the select permission will be combined in the following manner: 1. Limit - Minimum of the limits will be the limit of the inherited role 2. Allow aggregations - If any of the role allows aggregation, then the inherited role will allow aggregation 3. Scalar computed fields - same as table column fields, as in the above example APIs for inherited roles: ---------------------- 1. `add_inherited_role` `add_inherited_role` is the [metadata API](https://hasura.io/docs/1.0/graphql/core/api-reference/index.html#schema-metadata-api) to create a new inherited role. It accepts two arguments `role_name`: the name of the inherited role to be added (String) `role_set`: list of roles that need to be combined (Array of Strings) Example: ```json { "type": "add_inherited_role", "args": { "role_name":"combined_user", "role_set":[ "user", "user1" ] } } ``` After adding the inherited role, the inherited role can be used like single roles like earlier Note: An inherited role can only be created with non-inherited/singular roles. 2. `drop_inherited_role` The `drop_inherited_role` API accepts the name of the inherited role and drops it from the metadata. It accepts a single argument: `role_name`: name of the inherited role to be dropped Example: ```json { "type": "drop_inherited_role", "args": { "role_name":"combined_user" } } ``` Metadata --------- The derived roles metadata will be included under the `experimental_features` key while exporting the metadata. ```json { "experimental_features": { "derived_roles": [ { "role_name": "manager_is_employee_too", "role_set": [ "employee", "manager" ] } ] } } ``` Scope ------ Only postgres queries and subscriptions are supported in this PR. Important points: ----------------- 1. All columns exposed to an inherited role will be marked as `nullable`, this is done so that cell value nullification can be done. TODOs ------- - [ ] Tests - [ ] Test a GraphQL query running with a inherited role without enabling inherited roles in experimental features - [] Tests for aggregate queries, limit, computed fields, functions, subscriptions (?) - [ ] Introspection test with a inherited role (nullability changes in a inherited role) - [ ] Docs - [ ] Changelog Co-authored-by: Vamshi Surabhi <6562944+0x777@users.noreply.github.com> GitOrigin-RevId: 3b8ee1e11f5ceca80fe294f8c074d42fbccfec63
2021-03-08 14:14:13 +03:00
unset HASURA_GRAPHQL_INFER_FUNCTION_PERMISSIONS
unset HASURA_GRAPHQL_ADMIN_SECRET
[Preview] Inherited roles for postgres read queries fixes #3868 docker image - `hasura/graphql-engine:inherited-roles-preview-48b73a2de` Note: To be able to use the inherited roles feature, the graphql-engine should be started with the env variable `HASURA_GRAPHQL_EXPERIMENTAL_FEATURES` set to `inherited_roles`. Introduction ------------ This PR implements the idea of multiple roles as presented in this [paper](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/FGALanguageICDE07.pdf). The multiple roles feature in this PR can be used via inherited roles. An inherited role is a role which can be created by combining multiple singular roles. For example, if there are two roles `author` and `editor` configured in the graphql-engine, then we can create a inherited role with the name of `combined_author_editor` role which will combine the select permissions of the `author` and `editor` roles and then make GraphQL queries using the `combined_author_editor`. How are select permissions of different roles are combined? ------------------------------------------------------------ A select permission includes 5 things: 1. Columns accessible to the role 2. Row selection filter 3. Limit 4. Allow aggregation 5. Scalar computed fields accessible to the role Suppose there are two roles, `role1` gives access to the `address` column with row filter `P1` and `role2` gives access to both the `address` and the `phone` column with row filter `P2` and we create a new role `combined_roles` which combines `role1` and `role2`. Let's say the following GraphQL query is queried with the `combined_roles` role. ```graphql query { employees { address phone } } ``` This will translate to the following SQL query: ```sql select (case when (P1 or P2) then address else null end) as address, (case when P2 then phone else null end) as phone from employee where (P1 or P2) ``` The other parameters of the select permission will be combined in the following manner: 1. Limit - Minimum of the limits will be the limit of the inherited role 2. Allow aggregations - If any of the role allows aggregation, then the inherited role will allow aggregation 3. Scalar computed fields - same as table column fields, as in the above example APIs for inherited roles: ---------------------- 1. `add_inherited_role` `add_inherited_role` is the [metadata API](https://hasura.io/docs/1.0/graphql/core/api-reference/index.html#schema-metadata-api) to create a new inherited role. It accepts two arguments `role_name`: the name of the inherited role to be added (String) `role_set`: list of roles that need to be combined (Array of Strings) Example: ```json { "type": "add_inherited_role", "args": { "role_name":"combined_user", "role_set":[ "user", "user1" ] } } ``` After adding the inherited role, the inherited role can be used like single roles like earlier Note: An inherited role can only be created with non-inherited/singular roles. 2. `drop_inherited_role` The `drop_inherited_role` API accepts the name of the inherited role and drops it from the metadata. It accepts a single argument: `role_name`: name of the inherited role to be dropped Example: ```json { "type": "drop_inherited_role", "args": { "role_name":"combined_user" } } ``` Metadata --------- The derived roles metadata will be included under the `experimental_features` key while exporting the metadata. ```json { "experimental_features": { "derived_roles": [ { "role_name": "manager_is_employee_too", "role_set": [ "employee", "manager" ] } ] } } ``` Scope ------ Only postgres queries and subscriptions are supported in this PR. Important points: ----------------- 1. All columns exposed to an inherited role will be marked as `nullable`, this is done so that cell value nullification can be done. TODOs ------- - [ ] Tests - [ ] Test a GraphQL query running with a inherited role without enabling inherited roles in experimental features - [] Tests for aggregate queries, limit, computed fields, functions, subscriptions (?) - [ ] Introspection test with a inherited role (nullability changes in a inherited role) - [ ] Docs - [ ] Changelog Co-authored-by: Vamshi Surabhi <6562944+0x777@users.noreply.github.com> GitOrigin-RevId: 3b8ee1e11f5ceca80fe294f8c074d42fbccfec63
2021-03-08 14:14:13 +03:00
kill_hge_servers
;;
roles-inheritance)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH EXPERIMENTAL FEATURE: ROLES INHERITANCE ########>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_ENABLE_REMOTE_SCHEMA_PERMISSIONS="true"
export HASURA_GRAPHQL_INFER_FUNCTION_PERMISSIONS=false
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_roles_inheritance.py
unset HASURA_GRAPHQL_ADMIN_SECRET
unset HASURA_GRAPHQL_ENABLE_REMOTE_SCHEMA_PERMISSIONS
unset HASURA_GRAPHQL_INFER_FUNCTION_PERMISSIONS
kill_hge_servers
;;
naming-conventions)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH EXPERIMENTAL FEATURE: NAMING CONVENTIONS ########>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM"
run_hge_with_args serve
wait_for_port 8080
unset HASURA_GRAPHQL_EXPERIMENTAL_FEATURES
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_naming_conventions.py
kill_hge_servers
export HASURA_GRAPHQL_EXPERIMENTAL_FEATURES=naming_convention
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_naming_conventions.py
kill_hge_servers
# We are now going to test by setting the default naming convention to
# graphql-default. So now we don't need to set the naming convention in
# source customisation
export HASURA_GRAPHQL_DEFAULT_NAMING_CONVENTION="graphql-default"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_naming_conventions.py
unset HASURA_GRAPHQL_ADMIN_SECRET
export HASURA_GRAPHQL_EXPERIMENTAL_FEATURES=$DEFAULT_HASURA_EXPERIMENTAL_FEATURES
unset HASURA_GRAPHQL_DEFAULT_NAMING_CONVENTION
kill_hge_servers
;;
streaming-subscriptions)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH STREAMING SUBSCRIPTIONS #########################>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
run_hge_with_args serve
wait_for_port 8080
# run all the subscriptions tests with streaming subscriptions enabled
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_subscriptions.py
# run test `TestSubscriptionBasicNoAuth` in case of no authentication for Apollo-ws protocol
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_subscriptions.py::TestSubscriptionBasicNoAuth
unset HASURA_GRAPHQL_ADMIN_SECRET
unset HASURA_GRAPHQL_EXPERIMENTAL_FEATURES
kill_hge_servers
;;
query-caching)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE QUERY CACHING #####################################>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
# use only one capability to disable cache striping
run_hge_with_args +RTS -N1 -RTS serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_graphql_queries.py::TestGraphQLQueryCaching
kill_hge_servers
;;
query-logs)
# verbose logging tests
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH QUERY LOG ########>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_ENABLED_LOG_TYPES=" startup,http-log,webhook-log,websocket-log,query-log"
export HASURA_GRAPHQL_LOG_LEVEL="debug"
#run_hge_with_args serve
# we are doing this instead of calling run_hge_with_args, because we want to save in a custom log file
set -x
export LOGGING_TEST_LOGFILE_PATH="$OUTPUT_FOLDER/graphql-engine-verbose-logging.log"
"$GRAPHQL_ENGINE" serve >"$LOGGING_TEST_LOGFILE_PATH" 2>&1 &
HGE_PIDS=("${HGE_PIDS[@]}" $!)
set +x
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-logging \
test_logging.py
unset HASURA_GRAPHQL_ENABLED_LOG_TYPES
kill_hge_servers
# configurable jwk-refresh-log test
unset HASURA_GRAPHQL_AUTH_HOOK
unset HASURA_GRAPHQL_AUTH_HOOK_MODE
unset HASURA_GRAPHQL_JWT_SECRET
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH JWK REFRESH LOG ########> \n"
export JWK_SERVER_URL='http://localhost:5001'
# Start the JWK server.
# There is a fixture to do this, but when running in this fashion, we need to
# start the JWK server first so the HGE server can communicate with it.
python3 jwk_server.py >"$OUTPUT_FOLDER/configurable_log.log" 2>&1 &
JWKS_PID=$!
wait_for_port 5001
echo "Test: jwk-refresh-log type logs is enabled by default"
export HASURA_GRAPHQL_JWT_SECRET="{\"jwk_url\": \"${JWK_SERVER_URL}/jwk-cache-control?no-cache=true\"}"
#run_hge_with_args serve
# we are doing this instead of calling run_hge_with_args, because we want to save in a custom log file
set -x
export LOGGING_TEST_LOGFILE_PATH="$OUTPUT_FOLDER/graphql-engine-verbose-logging.log"
"$GRAPHQL_ENGINE" serve >"$LOGGING_TEST_LOGFILE_PATH" 2>&1 &
HGE_PIDS=("${HGE_PIDS[@]}" $!)
set +x
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-logging \
test_logging.py::TestConfiguragbleLogs
kill_hge_servers
echo "Test: no jwk-refresh-log type logs if it is not enabled"
export HASURA_GRAPHQL_ENABLED_LOG_TYPES=" startup,http-log,webhook-log,websocket-log,query-log"
#run_hge_with_args serve
# we are doing this instead of calling run_hge_with_args, because we want to save in a custom log file
set -x
export LOGGING_TEST_LOGFILE_PATH="$OUTPUT_FOLDER/graphql-engine-verbose-logging.log"
"$GRAPHQL_ENGINE" serve >"$LOGGING_TEST_LOGFILE_PATH" 2>&1 &
HGE_PIDS=("${HGE_PIDS[@]}" $!)
set +x
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-logging \
test_logging.py::TestConfiguragbleLogs
kill_hge_servers
unset HASURA_GRAPHQL_ENABLED_LOG_TYPES
unset HASURA_GRAPHQL_JWT_SECRET
unset JWK_SERVER_URL
kill $JWKS_PID
# end verbose logging tests
;;
startup-db-calls)
# verbose logging tests
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE STARTUP DB CALLS ########>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_ENABLED_LOG_TYPES=" startup,http-log,webhook-log,websocket-log,query-log"
export HASURA_GRAPHQL_LOG_LEVEL="debug"
#run_hge_with_args serve
# we are doing this instead of calling run_hge_with_args, because we want to save in a custom log file
set -x
export LOGGING_TEST_LOGFILE_PATH="$OUTPUT_FOLDER/graphql-engine-verbose-logging-db.log"
"$GRAPHQL_ENGINE" serve >"$LOGGING_TEST_LOGFILE_PATH" 2>&1 &
HGE_PIDS=("${HGE_PIDS[@]}" $!)
set +x
wait_for_port 8080
kill_hge_servers
# end verbose logging
# running HGE server again for pytest, the test will use the log generated from the previous run
# see https://github.com/hasura/graphql-engine-mono/pull/3813 for more information
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-startup-db-calls \
test_startup_db_calls.py
kill_hge_servers
unset HASURA_GRAPHQL_ENABLED_LOG_TYPES
unset HASURA_GRAPHQL_LOG_LEVEL
unset HASURA_GRAPHQL_ADMIN_SECRET
unset LOGGING_TEST_LOGFILE_PATH
;;
read-only-db)
## read-only DB tests; Hasura should start and run read queries against a read-only DB
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH READ-ONLY DATABASE ########>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_ENABLED_LOG_TYPES="startup,http-log,webhook-log,websocket-log,query-log"
export HASURA_GRAPHQL_LOG_LEVEL="debug"
export HASURA_GRAPHQL_DEV_MODE="false"
export HASURA_GRAPHQL_ADMIN_INTERNAL_ERRORS="false"
# setup the database for read-only access
# 'test_graphql_read_only_source.py' assumes 'HASURA_READONLY_DB_URL' is set
# Note: setting default_transaction_mode to read-only etc. doesn't work for
# DDL statements. To replicate read-only access even for DDLs, we need to
# create a read-only user
readonly_sql=$(cat <<EOF
CREATE USER hasuraro WITH PASSWORD 'passme';
GRANT CONNECT ON DATABASE pg_source_1 TO hasuraro;
GRANT USAGE ON SCHEMA public TO hasuraro;
GRANT SELECT ON ALL TABLES IN SCHEMA public TO hasuraro;
GRANT SELECT ON ALL TABLES IN SCHEMA pg_catalog TO hasuraro;
GRANT SELECT ON ALL TABLES IN SCHEMA information_schema TO hasuraro;
ALTER DEFAULT PRIVILEGES IN SCHEMA public GRANT SELECT ON TABLES TO hasuraro;
EOF
)
psql "$HASURA_GRAPHQL_PG_SOURCE_URL_1" -c "$readonly_sql"
PG_URL_HOST_PORT="$(sed -E 's#^postgresql://.+?@([^/]+)/.*#\1#' <<< "$HASURA_GRAPHQL_PG_SOURCE_URL_1")"
export HASURA_READONLY_DB_URL="postgresql://hasuraro:passme@${PG_URL_HOST_PORT}/pg_source_1"
run_hge_with_args serve
wait_for_port 8080
# and then test graphql queries work
pytest "${PYTEST_REPORTING_ARGS[@]}" \
--hge-urls "$HGE_URL" \
--pg-urls "$HASURA_GRAPHQL_PG_SOURCE_URL_1" \
--test-read-only-source \
test_graphql_read_only_source.py
unset HASURA_GRAPHQL_ENABLED_LOG_TYPES
kill_hge_servers
# end read-only DB tests
;;
remote-schema-https)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH SECURE REMOTE SCHEMA #########################>\n"
OLD_REMOTE_SCHEMAS_WEBHOOK_DOMAIN="${REMOTE_SCHEMAS_WEBHOOK_DOMAIN}"
export REMOTE_SCHEMAS_WEBHOOK_DOMAIN="https://localhost:5000"
init_ssl
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
--tls-ca-cert="$OUTPUT_FOLDER/ssl/ca.pem" --tls-ca-key="$OUTPUT_FOLDER/ssl/ca-key.pem" \
test_schema_stitching.py::TestRemoteSchemaBasic
export REMOTE_SCHEMAS_WEBHOOK_DOMAIN="${OLD_REMOTE_SCHEMAS_WEBHOOK_DOMAIN}"
kill_hge_servers
;;
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
webhook-request-context)
webhook_tests_check_root
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
echo -e "\n$(time_elapsed): <########## TEST WEBHOOK RECEIVES REQUEST DATA AS CONTEXT #########################>\n"
export HASURA_GRAPHQL_AUTH_HOOK="http://localhost:5594/"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="POST"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
run_hge_with_args serve
wait_for_port 8080
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_webhook_request_context.py
kill_hge_servers
;;
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
post-webhook)
webhook_tests_check_root
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET & WEBHOOK (POST) #########################>\n"
export HASURA_GRAPHQL_AUTH_HOOK="https://localhost:9090/"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="POST"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
init_ssl
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
--tls-ca-cert="$OUTPUT_FOLDER/ssl/ca.pem" --tls-ca-key="$OUTPUT_FOLDER/ssl/ca-key.pem" \
-- test_webhook.py
kill_hge_servers
;;
get-webhook)
webhook_tests_check_root
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET & WEBHOOK (GET) #########################>\n"
export HASURA_GRAPHQL_AUTH_HOOK="https://localhost:9090/"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="GET"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
init_ssl
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
run_hge_with_args serve
wait_for_port 8080
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
pytest "${PYTEST_COMMON_ARGS[@]}" \
--tls-ca-cert="$OUTPUT_FOLDER/ssl/ca.pem" --tls-ca-key="$OUTPUT_FOLDER/ssl/ca-key.pem" \
-- test_webhook.py
kill_hge_servers
;;
insecure-webhook)
webhook_tests_check_root
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN SECRET & HTTPS INSECURE WEBHOOK (GET) ########>\n"
export HASURA_GRAPHQL_AUTH_HOOK="https://localhost:9090/"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="GET"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
init_ssl
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
--tls-ca-cert="$OUTPUT_FOLDER/ssl/ca.pem" --tls-ca-key="$OUTPUT_FOLDER/ssl/ca-key.pem" \
test_webhook_insecure.py
kill_hge_servers
;;
insecure-webhook-with-admin-secret)
webhook_tests_check_root
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ADMIN_SECRET & HTTPS INSECURE WEBHOOK WITH ADMIN SECRET (POST) ########>\n"
export HASURA_GRAPHQL_AUTH_HOOK="https://localhost:9090/"
export HASURA_GRAPHQL_AUTH_HOOK_MODE="POST"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
init_ssl
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
server/tests-py: Start webhook.py inside the test harness. We use a helper service to start a webhook-based authentication service for some tests. This moves the initialization of the service out of _test-server.sh_ and into the Python test harness, as a fixture. In order to do this, I had to make a few changes. The main deviation is that we no longer run _all_ tests against an HGE with this authentication service, just a few (those in _test_webhook.py_). Because this reduced coverage, I have added some more tests there, which actually cover some areas not exacerbated elsewhere (mainly trying to use webhook credentials to talk to an admin-only endpoint). The webhook service can run both with and without TLS, and decide whether it's necessary to skip one of these based on the arguments passed and how HGE is started, according to the following logic: * If a TLS CA certificate is passed in, it will run with TLS, otherwise it will skip it. * If HGE was started externally and a TLS certificate is provided, it will skip running without TLS, as it will assume that HGE was configured to talk to a webhook over HTTPS. * Some tests should only be run with TLS; this is marked with a `tls_webhook_server` marker. * Some tests should only be run _without_ TLS; this is marked with a `no_tls_webhook_server` marker. The actual parameterization of the webhook service configuration is done through test subclasses, because normal pytest parameterization doesn't work with the `hge_fixture_env` hack that we use. Because `hge_fixture_env` is not a sanctioned way of conveying data between fixtures (and, unfortunately, there isn't a sanctioned way of doing this when the fixtures in question may not know about each other directly), parameterizing the `webhook_server` fixture doesn't actually parameterize `hge_server` properly. Subclassing forces this to work correctly. The certificate generation is moved to a Python fixture, so that we don't have to revoke the CA certificate for _test_webhook_insecure.py_; we can just generate a bogus certificate instead. The CA certificate is still generated in the _test-server.sh_ script, as it needs to be installed into the OS certificate store. Interestingly, the CA certificate installation wasn't actually working, because the certificates were written to the wrong location. This didn't cause any failures, as we weren't actually testing this behavior. This is now fixed with the other changes. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6363 GitOrigin-RevId: 0f277d374daa64f657257ed2a4c2057c74b911db
2022-10-20 21:58:36 +03:00
--tls-ca-cert="$OUTPUT_FOLDER/ssl/ca.pem" --tls-ca-key="$OUTPUT_FOLDER/ssl/ca-key.pem" \
test_webhook_insecure.py
kill_hge_servers
;;
apollo-federation)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH APOLLO FEDERATION ########>\n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM"
export HASURA_GRAPHQL_ENABLE_APOLLO_FEDERATION=true
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_apollo_federation.py
unset HASURA_GRAPHQL_ENABLE_APOLLO_FEDERATION
unset HASURA_GRAPHQL_ADMIN_SECRET
kill_hge_servers
;;
allowlist-queries)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH ALLOWLIST QUERIES ########> \n"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
export HASURA_GRAPHQL_ENABLE_ALLOWLIST=true
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_allowlist_queries.py
kill_hge_servers
;;
developer-api-tests)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH DEVELOPER API ENABLED ########>\n"
export HASURA_GRAPHQL_ENABLED_APIS="metadata,graphql,developer,config,pgdump"
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
run_hge_with_args serve --enabled-apis "$HASURA_GRAPHQL_ENABLED_APIS"
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
test_dev_endpoints.py
unset HASURA_GRAPHQL_ENABLED_APIS
kill_hge_servers
;;
jwk-url)
# TODO(swann): ditto, these have to be parallelised
# jwk test
unset HASURA_GRAPHQL_AUTH_HOOK
unset HASURA_GRAPHQL_AUTH_HOOK_MODE
unset HASURA_GRAPHQL_JWT_SECRET
export HASURA_GRAPHQL_ADMIN_SECRET="HGE$RANDOM$RANDOM"
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH JWK URL ########> \n"
export JWK_SERVER_URL='http://localhost:5001'
# Start the JWK server.
# There is a fixture to do this, but when running in this fashion, we need to
# start the JWK server first so the HGE server can communicate with it.
python3 jwk_server.py >"$OUTPUT_FOLDER/jwk_server.log" 2>&1 &
JWKS_PID=$!
wait_for_port 5001
echo "Test: Cache-Control with max-age=3"
export HASURA_GRAPHQL_JWT_SECRET="{\"jwk_url\": \"${JWK_SERVER_URL}/jwk-cache-control?max-age=3\"}"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
-- 'test_jwk.py::test_cache_control_header_max_age'
kill_hge_servers
unset HASURA_GRAPHQL_JWT_SECRET
echo "Test: Cache-Control with must-revalidate, max-age=3"
export HASURA_GRAPHQL_JWT_SECRET="{\"jwk_url\": \"${JWK_SERVER_URL}/jwk-cache-control?max-age=3&must-revalidate=true\"}"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
-- 'test_jwk.py::test_cache_control_header_max_age_must_revalidate'
kill_hge_servers
unset HASURA_GRAPHQL_JWT_SECRET
echo "Test: Cache-Control with must-revalidate"
export HASURA_GRAPHQL_JWT_SECRET="{\"jwk_url\": \"${JWK_SERVER_URL}/jwk-cache-control?must-revalidate=true\"}"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
-- 'test_jwk.py::test_cache_control_header_must_revalidate'
kill_hge_servers
unset HASURA_GRAPHQL_JWT_SECRET
echo "Test: Cache-Control with no-cache, public"
export HASURA_GRAPHQL_JWT_SECRET="{\"jwk_url\": \"${JWK_SERVER_URL}/jwk-cache-control?no-cache=true&public=true\"}"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
-- 'test_jwk.py::test_cache_control_header_no_cache_public'
kill_hge_servers
unset HASURA_GRAPHQL_JWT_SECRET
echo "Test: Cache-Control with no-store, max-age=3"
export HASURA_GRAPHQL_JWT_SECRET="{\"jwk_url\": \"${JWK_SERVER_URL}/jwk-cache-control?no-store=true&max-age=3\"}"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
-- 'test_jwk.py::test_cache_control_header_no_store_max_age'
kill_hge_servers
unset HASURA_GRAPHQL_JWT_SECRET
echo "Test: Expires with three second expiry"
export HASURA_GRAPHQL_JWT_SECRET="{\"jwk_url\": \"${JWK_SERVER_URL}/jwk-expires?seconds=3\"}"
run_hge_with_args serve
wait_for_port 8080
pytest "${PYTEST_COMMON_ARGS[@]}" \
-- 'test_jwk.py::test_expires_header'
kill_hge_servers
unset HASURA_GRAPHQL_JWT_SECRET
kill $JWKS_PID
# end jwk url test
;;
horizontal-scaling)
# horizontal scale test
unset HASURA_GRAPHQL_AUTH_HOOK
unset HASURA_GRAPHQL_AUTH_HOOK_MODE
unset HASURA_GRAPHQL_ADMIN_SECRET
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH HORIZONTAL SCALING ########>\n"
HASURA_HS_TEST_DB='postgresql://postgres:postgres@localhost:6543/hs_hge_test'
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
if ! psql "$HASURA_GRAPHQL_DATABASE_URL" -c "SELECT 1 FROM pg_database WHERE datname = 'hs_hge_test'" | grep -q -F '(1 row)'; then
psql "$HASURA_GRAPHQL_DATABASE_URL" -c 'CREATE DATABASE hs_hge_test;'
fi
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
pgUserInfo=$(python3 -c '
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
import os
from urllib.parse import urlparse
uri = urlparse( os.environ["HASURA_GRAPHQL_DATABASE_URL"] )
if uri.password:
print("password="+uri.password+" user="+uri.username)
else:
print("user="+uri.username)')
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
pgDbInfo=$(psql "$HASURA_GRAPHQL_DATABASE_URL" -c "SELECT concat(' host=',inet_server_addr(),' port=', inet_server_port(),' dbname=',current_database())" | sed -n '3 p')
# create pgbouncer user
id pgbouncer || useradd pgbouncer
cd "$CIRCLECI_FOLDER"
mkdir -p pgbouncer
chown -R pgbouncer:pgbouncer pgbouncer
echo '[databases]
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
hs_hge_test = '"$pgDbInfo" "$pgUserInfo"'
[pgbouncer]
listen_port = 6543
listen_addr = 127.0.0.1
logfile = pgbouncer/pgbouncer.log
pidfile = pgbouncer/pgbouncer.pid
auth_type = md5
auth_file = pgbouncer/users.txt
admin_users = postgres' >pgbouncer/pgbouncer.ini
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
# start pgbouncer
pgbouncer -u pgbouncer -d pgbouncer/pgbouncer.ini
cd "$PYTEST_ROOT"
sleep 2
# start 1st server
run_hge_with_args --database-url "$HASURA_HS_TEST_DB" serve
wait_for_port 8080
# start 2nd server
run_hge_with_args --database-url "$HASURA_HS_TEST_DB" serve \
--server-port 8081
wait_for_port 8081
# run test
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-hge-scale-url="http://localhost:8081" \
test_horizontal_scale.py
# Shutdown pgbouncer
psql "postgresql://postgres:postgres@localhost:6543/pgbouncer" -c "SHUTDOWN;" || true
cd "$CIRCLECI_FOLDER"
# start pgbouncer again
pgbouncer -u pgbouncer -d pgbouncer/pgbouncer.ini
cd "$PYTEST_ROOT"
# sleep for 20 seconds
sleep 20
# run test
pytest "${PYTEST_COMMON_ARGS[@]}" \
--test-hge-scale-url="http://localhost:8081" \
test_horizontal_scale.py
# Shutdown pgbouncer
psql "postgresql://postgres:postgres@localhost:6543/pgbouncer" -c "SHUTDOWN;" || true
kill_hge_servers
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
psql "$HASURA_GRAPHQL_DATABASE_URL" -c "drop database hs_hge_test;"
sleep 4
unset HASURA_HS_TEST_DB
# end horizontal scale test
;;
#
# ###########################################
# the following backend-* tests are excluded from `server-test-names.txt`
# and are run via their respective `test_oss_server_*` jobs
#
# [Specifying Pytests with -k flag]
# tests are run with the -k flag to filter on common and
# backend-specific test classes using keyword expressions.
#
# this reduces the number of unrelated tests skipped, which
# avoids an increasingly negative impact on our test run
# time as we add more backends and tests.
#
# https://docs..org/en/6.2.x/usage.html#specifying-tests-selecting-tests
# https://github.com/hasura/graphql-engine/blob/master/server/py-tests/README.md#running-bigquery-tests
#
backend-mssql)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH SQL SERVER BACKEND ###########################################>\n"
export HASURA_GRAPHQL_EXPERIMENTAL_FEATURES=naming_convention
run_hge_with_args serve
wait_for_port 8080
source_data_sources_utils
add_mssql_source 8080 "$HASURA_GRAPHQL_MSSQL_SOURCE_URL"
# See note [Specifying Pytests with -k flag]
pytest "${PYTEST_COMMON_ARGS[@]}" \
--backend=mssql \
-k "MSSQL"
# start inherited roles test
echo -e "\n$(time_elapsed): <########## TEST INHERITED-ROLES WITH SQL SERVER BACKEND ###########################################>\n"
pytest "${PYTEST_COMMON_ARGS[@]}" \
--backend=mssql \
-k TestGraphQLInheritedRolesMSSQL
# end inherited roles test
# start naming conventions test (failure for other than postgres backend)
echo -e "\n$(time_elapsed): <########## TEST NAMING CONVENTIONS WITH SQL SERVER BACKEND ###########################################>\n"
pytest "${PYTEST_COMMON_ARGS[@]}" \
--backend mssql \
-k TestNamingConventionsFailure
export HASURA_GRAPHQL_EXPERIMENTAL_FEATURES=$DEFAULT_HASURA_EXPERIMENTAL_FEATURES
# end naming conventions test
kill_hge_servers
;;
backend-citus)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH CITUS BACKEND ###########################################>\n"
run_hge_with_args serve
wait_for_port 8080
source_data_sources_utils
add_citus_source 8080 "$HASURA_GRAPHQL_CITUS_SOURCE_URL"
# See note [Specifying Pytests with -k flag]
pytest "${PYTEST_COMMON_ARGS[@]}" \
--backend=citus \
-k "Citus"
kill_hge_servers
;;
backend-bigquery)
echo -e "\n$(time_elapsed): <########## TEST GRAPHQL-ENGINE WITH BIGQUERY BACKEND ###########################################>\n"
# shellcheck source=../scripts/bigquery.sh
server/tests: ephemeral BigQuery projects for CI test jobs _Problem_ We currently run teardown/`DELETE` statements on the same, shared `hasura_test` dataset. This is not ideal as parallel test runs can and do affect each other, resulting in nondeterministic CI failures. Closes https://github.com/hasura/graphql-engine-mono/issues/2521 _Solution and design_ This PR introduces ephemeral, isolated projects for each test run _in CI only_. Projects are created within [the Google Cloud Platform `data-sources-test-bigquery` directory](https://console.cloud.google.com/iam-admin/settings?folder=704256416468&orgonly=true&supportedpurview=organizationId) on each test run, and destroyed afterwards. I've only introduced this change in CI for the time being: 1. this isn't as much of an issue locally because we're less likely to run bigquery tests in parallel. 2. to more quickly unblock https://github.com/hasura/graphql-engine/issues/7929. 3. to limit the number of new projects created until we have a better idea of our usage vs GCP quota/limits. Also updated the [internal wiki here](https://github.com/hasura/graphql-engine-mono/wiki/Testing-BigQuery) with this info. _To verify_ - CI: [this job](https://buildkite.com/hasura/graphql-engine-mono/builds/3770#89e5bac6-16fe-447e-bcda-85cd47ea1b77) successfully runs all tests on a temporary project & dataset - local: follow [these steps](https://github.com/hasura/graphql-engine-mono/wiki/Testing-BigQuery#ci--optional-dedicated-gcp-project-for-tests) to try the same setup locally PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3240 GitOrigin-RevId: d88d9cb7922266bfa962cfcb481e0272b8929a5d
2022-01-25 11:26:32 +03:00
source "$CIRCLECI_FOLDER/../scripts/bigquery.sh" && verify_bigquery_pytest_env
run_hge_with_args serve
wait_for_port 8080
server/tests: ephemeral BigQuery projects for CI test jobs _Problem_ We currently run teardown/`DELETE` statements on the same, shared `hasura_test` dataset. This is not ideal as parallel test runs can and do affect each other, resulting in nondeterministic CI failures. Closes https://github.com/hasura/graphql-engine-mono/issues/2521 _Solution and design_ This PR introduces ephemeral, isolated projects for each test run _in CI only_. Projects are created within [the Google Cloud Platform `data-sources-test-bigquery` directory](https://console.cloud.google.com/iam-admin/settings?folder=704256416468&orgonly=true&supportedpurview=organizationId) on each test run, and destroyed afterwards. I've only introduced this change in CI for the time being: 1. this isn't as much of an issue locally because we're less likely to run bigquery tests in parallel. 2. to more quickly unblock https://github.com/hasura/graphql-engine/issues/7929. 3. to limit the number of new projects created until we have a better idea of our usage vs GCP quota/limits. Also updated the [internal wiki here](https://github.com/hasura/graphql-engine-mono/wiki/Testing-BigQuery) with this info. _To verify_ - CI: [this job](https://buildkite.com/hasura/graphql-engine-mono/builds/3770#89e5bac6-16fe-447e-bcda-85cd47ea1b77) successfully runs all tests on a temporary project & dataset - local: follow [these steps](https://github.com/hasura/graphql-engine-mono/wiki/Testing-BigQuery#ci--optional-dedicated-gcp-project-for-tests) to try the same setup locally PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3240 GitOrigin-RevId: d88d9cb7922266bfa962cfcb481e0272b8929a5d
2022-01-25 11:26:32 +03:00
source_data_sources_utils
add_bigquery_source 8080
# See note [Specifying Pytests with -k flag]
pytest "${PYTEST_COMMON_ARGS[@]}" \
--backend=bigquery \
-k "Bigquery"
kill_hge_servers
;;
esac
echo "Finished running tests on node $CIRCLE_NODE_INDEX of $CIRCLE_NODE_TOTAL"
run graphql tests on both http and websocket; add parallelism (close #1868) (#1921) Examples 1) ` pytest --hge-urls "http://127.0.0.1:8080" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" -vv ` 2) `pytest --hge-urls "http://127.0.0.1:8080" "http://127.0.0.1:8081" --pg-urls "postgresql://admin@127.0.0.1:5432/hge_tests" "postgresql://admin@127.0.0.1:5432/hge_tests2" -vv ` ### Solution and Design <!-- How is this issue solved/fixed? What is the design? --> <!-- It's better if we elaborate --> #### Reducing execution time of tests - The Schema setup and teardown, which were earlier done per test method, usually takes around 1 sec. - For mutations, the model has now been changed to only do schema setup and teardown once per test class. - A data setup and teardown will be done once per test instead (usually takes ~10ms). - For the test class to get this behaviour, one can can extend the class `DefaultTestMutations`. - The function `dir()` should be define which returns the location of the configuration folder. - Inside the configuration folder, there should be - Files `<conf_dir>/schema_setup.yaml` and `<conf_dir>/schema_teardown.yaml`, which has the metadata query executed during schema setup and teardown respectively - Files named `<conf_dir>/values_setup.yaml` and `<conf_dir>/values_teardown.yaml`. These files are executed to setup and remove data from the tables respectively. #### Running Graphql queries on both http and websockets - Each GraphQL query/mutation is run on the both HTTP and websocket protocols - Pytests test parameterisation is used to achieve this - The errors over websockets are slightly different from that on HTTP - The code takes care of converting the errors in HTTP to errors in websockets #### Parallel executation of tests. - The plugin pytest-xdist helps in running tests on parallel workers. - We are using this plugin to group tests by file and run on different workers. - Parallel test worker processes operate on separate postgres databases(and separate graphql-engines connected to these databases). Thus tests on one worker will not affect the tests on the other worker. - With two workers, this decreases execution times by half, as the tests on event triggers usually takes a long time, but does not consume much CPU.
2019-04-08 10:22:38 +03:00
echo -e "\n$(time_elapsed): <########## DONE ########>\n"