graphql-engine/server/bench-wrk/run_hge.py
nizar-m f8a7312a30
Regression benchmarks setup (#3310)
* Benchmark GraphQL queries using wrk

* fix console assets dir

* Store wrk parameters as well

* Add details about storing results in Readme

* Remove files in bench-wrk while computing server shasum

* Instead of just getting maximum throughput per query per version,
create plots using wrk2 for a given set of requests per second.
The maximum throughput is used to see what values of requests per second are feasible.

* Add id for version dropdown

* Allow specifiying env and args for GraphQL Engine

1) Arguments defined after -- will be applied as arguments to Hasura GraphQL Engine
2) Script will also pass the environmental variables to Hasura GraphQL Engine instances

Hasura GraphQL engine can be run with the given environmental variables and arguments as follows

$ export HASURA_GRAPHQL_...=....
$ python3 hge_wrk_bench.py -- --hge_arg1 val1 --hge_arg2 val2 ...

* Use matplotlib instead of plotly for figures

* Show throughput graph also.

It maybe useful in checking performance regression across versions

* Support storing results in s3

Use --upload-root-uri 's3://bucket/path' to upload results inside the
given path.When specified, the results will be uploaded to the bucket,
including latencies, latency histogram, and the test setup info.
The s3 credentials should be provided as given in AWS boto3 documentation.

* Allow specifying a name for the test scenario

* Fix open latency uri bug

* Update wrk docker image

* Keep ylim a little higher than maximum so that the throughput plot is clearly visible

* Show throughput plots for multiple queries at the same time

* 1) Adjust size of dropdowns
2) Make label for requests/sec invisible when plot type is throughput

* 1) Adding boto3 to requirements.txt
2) Removing CPU Key print line
3) Adding info about the tests that will be run with wrk2

* Docker builder fo wrk-websocket-server

* Make it optional to setup remote graphql-engine

* Listen on all interfaces and enable ping thread

* Add bench_scripts to wrk-websocket-server docker

* Use 127.0.0.1 instead of 'localhost' to address local hge

For some reason it seems wrk was hanging trying to resolve 'localhost'.
ping was able to fine from the same container, so I'm not sure what the
deal was. Probably some local misconfiguration on my machine, but maybe
this change will also help others.

* Store latency samples in subdirectory, server_shasum just once at start, additional docs

* Add a note on running the benchmarks in the simplest way

* Add a new section on how to run benchmarks on a new linux hosted instance

Co-authored-by: Nizar Malangadan <nizar-m@users.noreply.github.com>
Co-authored-by: Brandon Simmons <brandon.m.simmons@gmail.com>
Co-authored-by: Karthikeyan Chinnakonda <karthikeyan@hasura.io>
Co-authored-by: Brandon Simmons <brandon@hasura.io>
Co-authored-by: Vamshi Surabhi <0x777@users.noreply.github.com>
2020-06-19 22:40:17 +05:30

465 lines
16 KiB
Python

import os
import subprocess
import argparse
import json
import signal
import time
import contextlib
import requests
import inflection
import docker
from colorama import Fore, Style
def rm_file_if_exists(f):
"""Remove a file if it exists"""
with contextlib.suppress(FileNotFoundError):
os.remove(f)
class HGEError(Exception):
"""Exception type for class HGE"""
class HGE:
default_graphql_env = {
'HASURA_GRAPHQL_ENABLE_TELEMETRY': 'false',
'EVENT_WEBHOOK_HEADER': "MyEnvValue",
'HASURA_GRAPHQL_STRINGIFY_NUMERIC_TYPES': 'true',
'HASURA_GRAPHQL_CONSOLE_ASSETS_DIR' : '../../console/static/dist/',
'HASURA_GRAPHQL_ENABLE_CONSOLE' : 'true'
}
def __init__(self, pg, port_allocator, docker_image=None, log_file='hge.log', url=None, args=[]):
self.pg = pg
self.log_file = log_file
if self.log_file:
self.tix_file = self.log_file[:-4] + '.tix'
self.docker_image = docker_image
self.introspection = None
self.obj_fk_rels = set()
self.arr_fk_rels = set()
self.port_allocator = port_allocator
self.url = url
self.proc = None
self.container = None
self.args = args
def admin_secret(self):
admin_secret_env = os.environ.get('HASURA_GRAPHQL_ADMIN_SECRET')
parser = argparse.ArgumentParser()
parser.add_argument('--admin-secret', metavar='HASURA_GRAPHQL_ADMIN_SECRET', required=False)
admin_secret_arg = parser.parse_known_args(self.args)[0].admin_secret
return admin_secret_arg or admin_secret_env
@classmethod
def do_stack_build(cls):
print(Fore.YELLOW + "Performing Stack build first" + Style.RESET_ALL)
# 'stack run' below will also build, but we want to make sure that's a
# noop so the server starts right away
subprocess.check_call( ['cabal', 'new-build', 'exe:graphql-engine'])
def get_hge_env(self):
hge_env = {
**os.environ,
**self.default_graphql_env.copy(),
'HASURA_GRAPHQL_DATABASE_URL': self.pg.url,
'HASURA_GRAPHQL_SERVER_PORT': str(self.port),
'HASURA_GRAPHQL_SERVER_HOST': '127.0.0.1',
'HPCTIXFILE' : self.tix_file
}
return hge_env
def run(self):
if self.url:
return
if self.docker_image:
self.run_with_docker()
else:
self.run_with_cabal()
def run_with_docker(self):
if self.url:
return
self.port = self.port_allocator.allocate_port(8080)
hge_env = self.get_hge_env()
process_args = ['graphql-engine', 'serve', *self.args]
docker_ports = {str(self.port) + '/tcp': ('127.0.0.1', self.port)}
self.docker_client = docker.from_env()
print("Running GraphQL Engine docker with image:",
self.docker_image, '(port:{})'.format(self.port))
print(process_args)
self.container = self.docker_client.containers.run(
self.docker_image,
command=process_args,
detach=True,
ports=docker_ports,
environment=hge_env,
network_mode='host',
volumes={}
)
self.url = 'http://127.0.0.1:' + str(self.port)
print("Waiting for GraphQL Engine to be running.", end='')
self.wait_for_start()
def run_with_cabal(self):
if self.url:
return
self.port = self.port_allocator.allocate_port(8080)
rm_file_if_exists(self.tix_file)
hge_env = self.get_hge_env()
process_args = ['cabal', 'new-run', '--', 'exe:graphql-engine', 'serve', *self.args]
print("Running GraphQL with 'cabal run': (port:{})".format(self.port))
print(process_args)
self.log_fp = open(self.log_file, 'w')
self.proc = subprocess.Popen(
process_args,
env=hge_env,
shell=False,
bufsize=-1,
start_new_session=True,
stdout=self.log_fp,
stderr=subprocess.STDOUT
)
self.url = 'http://127.0.0.1:' + str(self.port)
print("Waiting for GraphQL Engine to be running.", end='')
self.wait_for_start()
def check_if_process_is_running(self):
if self.proc.poll() is not None:
with open(self.log_file) as fr:
raise HGEError(
"GraphQL engine failed with error: " + fr.read())
def check_if_container_is_running(self):
self.container.reload()
if self.container.status == 'exited':
raise HGEError(
"GraphQL engine failed with error: \n" +
self.container.logs(stdout=True, stderr=True).decode('ascii')
)
def wait_for_start(self, timeout=120):
if timeout <= 0:
raise HGEError("Timeout waiting for graphql process to start")
if self.proc:
self.check_if_process_is_running()
elif self.container:
self.check_if_container_is_running()
try:
q = { 'query': 'query { __typename }' }
r = requests.post(self.url + '/v1/graphql',json.dumps(q),headers=self.admin_auth_headers())
if r.status_code == 200:
print()
return
except requests.exceptions.ConnectionError:
pass
except ConnectionError:
pass
print(".", end="", flush=True),
sleep_time = 0.5
time.sleep(sleep_time)
self.wait_for_start(timeout - sleep_time)
def teardown(self):
if getattr(self, 'log_fp', None):
self.log_fp.close()
self.log_fp = None
if self.proc:
self.cleanup_process()
elif self.container:
self.cleanup_docker()
def cleanup_process(self):
# TODO hangs
print(Fore.YELLOW + "Stopping graphql engine at port:", self.port, Style.RESET_ALL)
pgrp = os.getpgid(self.proc.pid)
os.killpg(pgrp, signal.SIGTERM)
# NOTE this doesn't seem to work, although a SIGINT from terminal does ...
# self.proc.send_signal(signal.SIGINT)
self.proc.wait()
self.proc = None
def cleanup_docker(self):
cntnr_info = "HGE docker container " + self.container.name + " " + repr(self.container.image)
print(Fore.YELLOW + "Stopping " + cntnr_info + Style.RESET_ALL)
self.container.stop()
print(Fore.YELLOW + "Removing " + cntnr_info + Style.RESET_ALL)
self.container.remove()
self.container = None
def admin_auth_headers(self):
headers = {}
if self.admin_secret():
headers['X-Hasura-Admin-Secret'] = self.admin_secret()
return headers
def v1q(self, q, exp_status=200):
resp = requests.post(self.url + '/v1/query', json.dumps(q), headers=self.admin_auth_headers())
assert resp.status_code == exp_status, (resp.status_code, resp.json())
return resp.json()
def graphql_q(self, query, variables={}, exp_status = 200):
q = {'query': query}
if variables:
q['variables'] = variables
resp = requests.post(self.url + '/v1/graphql', json.dumps(q), headers=self.admin_auth_headers())
assert resp.status_code == exp_status, (resp.status_code, resp.json())
assert 'errors' not in resp.json(), resp.json()
return resp.json()
def track_all_tables_in_schema(self, schema='public'):
print("Track all tables in schema ", schema)
all_tables = self.pg.get_all_tables_in_a_schema(schema)
all_tables = [ {'schema': schema, 'name': t}
for t in all_tables ]
return self.track_tables(all_tables)
def run_bulk(self, queries, exp_status = 200):
bulk_q = {
'type': 'bulk',
'args': queries
}
return self.v1q(bulk_q, exp_status)
def select_simple(self, table, columns):
query = {
'type': 'select',
'args': {
'table': table,
'columns': columns
}
}
return self.v1q(query)
def get_all_tracked_tables(self):
table = {
'schema': 'hdb_catalog',
'name': 'hdb_table'
}
columns = ['table_schema', 'table_name']
resp = self.select_simple(table, columns)
tables = []
for row in resp:
tables.append({
'schema': row['table_schema'],
'name': row['table_name']
})
return tables
def track_tables(self, tables, exp_status=200):
queries = []
for table in tables:
q = {
'type' : 'track_table',
'args' : table
}
queries.append(q)
return self.run_bulk(queries, exp_status)
def track_table(self, table, exp_status=200):
q = self.mk_track_table_q(table)
return self.v1q(q, exp_status)
def mk_track_table_q(self, table):
return {
'type' : 'track_table',
'args' : table
}
def add_remote_schema(self, name, remote_url, headers={}, client_hdrs=False):
def hdr_name_val_pair(headers):
nvp = []
for (k,v) in headers.items():
nvp.append({'name': k, 'value': v})
return nvp
if len(headers) > 0:
client_hdrs = True
q = {
'type' : 'add_remote_schema',
'args': {
'name': name,
'comment': name,
'definition': {
'url': remote_url,
'headers': hdr_name_val_pair(headers),
'forward_client_headers': client_hdrs
}
}
}
return self.v1q(q)
def create_remote_obj_rel_to_itself(self, tables_schema, remote, remote_tables_schema):
print("Creating remote relationship to the tables in schema {} to itself using remote {}".format(tables_schema, remote))
fk_constrnts = self.pg.get_all_fk_constraints(tables_schema)
for (s, _, t, c, _, ft, _) in fk_constrnts:
table_cols = self.pg.get_all_columns_of_a_table(t, s)
if not 'id' in table_cols:
continue
rel_name = 'remote_' + inflection.singularize(t) + '_via_' + c
query ={
'type': 'create_remote_relationship',
'args' : {
'name' : rel_name,
'table' : {
'schema': s,
'name': t
},
'remote_schema': remote,
'hasura_fields': ['id', c],
'remote_field': {
remote_tables_schema + '_' + ft + '_by_pk' : {
'arguments': {
'id': '$' + c
},
'field': {
inflection.pluralize(t) + '_by_' + c: {
'arguments' : {
'where': {
c : {
'_eq': '$id'
}
}
}
}
}
}
}
}
}
print(query)
self.v1q(query)
def create_remote_obj_fk_ish_relationships(self, tables_schema, remote, remote_tables_schema):
print("Creating object foreign key ish relationships for tables in schema {} using remote {}".format(tables_schema, remote))
fk_constrnts = self.pg.get_all_fk_constraints(tables_schema)
for (s, _, t, c, _, ft, _) in fk_constrnts:
rel_name = inflection.singularize(ft)
if c.endswith('_id'):
rel_name = c[:-3]
rel_name = 'remote_' + rel_name
query ={
'type': 'create_remote_relationship',
'args' : {
'name' : rel_name,
'table' : {
'schema': s,
'name': t
},
'remote_schema': remote,
'hasura_fields': [c],
'remote_field': {
remote_tables_schema + '_' + ft + '_by_pk' : {
'arguments' : {
'id': '$' + c
}
}
}
}
}
print(query)
self.v1q(query)
def create_obj_fk_relationships(self, schema='public'):
print("Creating object foreign key relationships for tables in schema ", schema)
fk_constrnts = self.pg.get_all_fk_constraints(schema)
queries = []
for (s, _, t, c, _, ft, _) in fk_constrnts:
rel_name = inflection.singularize(ft)
if c.endswith('_id'):
rel_name = c[:-3]
table_cols = self.pg.get_all_columns_of_a_table(t, s)
if rel_name in table_cols:
rel_name += '_' + inflection.singularize(ft)
queries.append({
'type' : 'create_object_relationship',
'args': {
'table': {
'schema': s,
'name': t
},
'name': rel_name,
'using': {
'foreign_key_constraint_on': c
}
}
})
self.obj_fk_rels.add(((s,t),rel_name))
return self.run_bulk(queries)
def create_remote_arr_fk_ish_relationships(self, tables_schema, remote, remote_tables_schema):
fk_constrnts = self.pg.get_all_fk_constraints(tables_schema)
for (_, _, t, c, fs, ft, _) in fk_constrnts:
rel_name = 'remote_' + inflection.pluralize(t) + '_by_' + c
query ={
'type': 'create_remote_relationship',
'args' : {
'name' : rel_name,
'table' : {
'schema': fs,
'name': ft
},
'remote_schema': remote,
'hasura_fields': ['id'],
'remote_field': {
remote_tables_schema + '_' + t : {
'arguments' : {
'where': {
c : {
'_eq': '$id'
}
}
}
}
}
}
}
print(query)
self.v1q(query)
def create_arr_fk_relationships(self, schema='public'):
print("Creating array foreign key relationships for tables in schema ", schema)
fk_constrnts = self.pg.get_all_fk_constraints(schema)
queries = []
for (s, _, t, c, fs, ft, _) in fk_constrnts:
rel_name = inflection.pluralize(t) + '_by_' + c
queries.append({
'type' : 'create_array_relationship',
'args': {
'table': {
'schema': fs,
'name': ft
},
'name': rel_name,
'using': {
'foreign_key_constraint_on': {
'table': {
'schema': s,
'name': t
},
'column': c
}
}
}
})
self.arr_fk_rels.add(((fs,ft),rel_name))
return self.run_bulk(queries)
def run_sql(self, sql):
"""Run given SQL query"""
def mk_run_sql_q(sql):
return {
'type' : 'run_sql',
'args': {
'sql' : sql
}
}
return self.v1q(mk_run_sql_q(sql))