## Description
### I want to speak to the `Manager`
Oh boy. This PR is both fairly straightforward and overreaching, so let's break it down.
For most network access, we need a [`HTTP.Manager`](https://hackage.haskell.org/package/http-client-0.1.0.0/docs/Network-HTTP-Client-Manager.html). It is created only once, at the top level, when starting the engine, and is then threaded through the application to wherever we need to make a network call. As of main, the way we do this is not standardized: most of the GraphQL execution code passes it "manually" as a function argument throughout the code. We also have a custom monad constraint, `HasHttpManagerM`, that describes a monad's ability to provide a manager. And, finally, several parts of the code store the manager in some kind of argument structure, such as `RunT`'s `RunCtx`.
This PR's first goal is to harmonize all of this: we always create the manager at the root, and we already have it when we do our very first `runReaderT`. Wouldn't it make sense for the rest of the code to not manually pass it anywhere, to not store it anywhere, but to always rely on the current monad providing it? This is, in short, what this PR does: it implements a constraint on the base monads, so that they provide the manager, and removes most explicit passing from the code.
### First come, first served
One way this PR goes a tiny bit further than "just" doing the aforementioned harmonization is that it starts the process of implementing the "Services oriented architecture" roughly outlined in this [draft document](https://docs.google.com/document/d/1FAigqrST0juU1WcT4HIxJxe1iEBwTuBZodTaeUvsKqQ/edit?usp=sharing). Instead of using the existing `HasHTTPManagerM`, this PR revamps it into the `ProvidesNetwork` service.
The idea is, again, that we should make all "external" dependencies of the engine, all things that the core of the engine doesn't care about, a "service". This allows us to define clear APIs for features, to choose different implementations based on which version of the engine we're running, harmonizes our many scattered monadic constraints... Which is why this service is called "Network": we can refine it, moving forward, to be the constraint that defines how all network communication is to operate, instead of relying on disparate classes constraint or hardcoded decisions. A comment in the code clarifies this intent.
### Side-effects? In my Haskell?
This PR also unavoidably touches some other aspects of the codebase. One such example: it introduces `Hasura.App.AppContext`, named after `HasuraPro.Context.AppContext`: a name for the reader structure at the base level. It also transforms `Handler` from a type alias to a newtype, as `Handler` is where we actually enforce HTTP limits; but without `Handler` being a distinct type, any code path could simply do a `runExceptT $ runReader` and forget to enforce them.
(As a rule of thumb, i am starting to consider any straggling `runReaderT` or `runExceptT` as a code smell: we should not stack / unstack monads haphazardly, and every layer should be an opaque `newtype` with a corresponding run function.)
## Further work
In several places, i have left TODOs when i have encountered things that suggest that we should do further unrelated cleanups. I'll write down the follow-up steps, either in the aforementioned document or on slack. But, in short, at a glance, in approximate order, we could:
- delete `ExecutionCtx` as it is only a subset of `ServerCtx`, and remove one more `runReaderT` call
- delete `ServerConfigCtx` as it is only a subset of `ServerCtx`, and remove it from `RunCtx`
- remove `ServerCtx` from `HandlerCtx`, and make it part of `AppContext`, or even make it the `AppContext` altogether (since, at least for the OSS version, `AppContext` is there again only a subset)
- remove `CacheBuildParams` and `CacheBuild` altogether, as they're just a distinct stack that is a `ReaderT` on top of `IO` that contains, you guessed it, the same thing as `ServerCtx`
- move `RunT` out of `RQL.Types` and rename it, since after the previous cleanups **it only contains `UserInfo`**; it could be bundled with the authentication service, made a small implementation detail in `Hasura.Server.Auth`
- rename `PGMetadaStorageT` to something a bit more accurate, such as `App`, and enforce its IO base
This would significantly simply our complex stack. From there, or in parallel, we can start moving existing dependencies as Services. For the purpose of supporting read replicas entitlement, we could move `MonadResolveSource` to a `SourceResolver` service, as attempted in #7653, and transform `UserAuthenticationM` into a `Authentication` service.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7736
GitOrigin-RevId: 68cce710eb9e7d752bda1ba0c49541d24df8209f
- Inline a few instances to avoid code duplication
- Use `(<$>)` to avoid `let`
- Improve error reporting when types of invalid kind are specified in `possibleTypes` or `interfaces`.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7540
GitOrigin-RevId: 954fb710f94a275daff938b9a6e58765c4286d0c
### Description
Each Backend executes queries against the database in a slightly different stack: Postgres uses its own `TXeT`, MSSQL uses a variant of it, BigQuery is simply in `ExceptT QErr IO`... To accommodate those variations, we had originally introduced an `ExecutionMonad b` type family in `BackendExecute`, allowing each backend to describe its own stack. It was then up to that backend's `BackendTransport` instance to implement running said stack, and converting the result back into our main app monad.
However, this was not without complications: `TraceT` is one of them: as it usually needs to be on the top of the stack, converting from one stack to the other implies the use `interpTraceT`, which is quite monstrous. Furthermore, as part of the Entitlement Services work, we're trying to move to a "Services" architecture in which the entire engine runs in one base monad, that delegates features and dependencies to monad constraints; and as a result we'd like to minimize the number of different monad stacks we have to maintain and translate from and to in the codebase.
To improve things, this PR changes `ExecutionMonad b` from an _absolute_ stack to a _relative_ one: i.e.: what needs to be stacked on top of our base monad for the execution. In `Transport`, we then only need to pop the top of the stack, and voila. This greatly simplifies the implementation of the backends, as there's no longer any need to do any stack transformation: MySQL's implementation becomes a `runIdentityT`! This also removes most mentions of `TraceT` from the execution code since it's no longer required: we can rely on the base monad's existing `MonadTrace` constraint.
To continue encapsulating monadic actions in `DBStepInfo` and avoid threading a bunch of `forall` all over the place, this PR introduces a small local helper: `OnBaseMonad`. One only downside of all this is that this requires adding `MonadBaseControl IO m` constraint all over the place: previously, we would run directly on `IO` and lift, and would therefore not need to bring that constraint all the way.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7789
GitOrigin-RevId: e9b2e431c5c47fa9851abf87545c0415ff6d1a12
Basic MongoDB agent. This is intended as a starting point for playing with nested documents in a MongoDB back end. Currently supports basic queries with projections, where expressions, limit and offset. No support for joins, aggregates or mutations.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7840
GitOrigin-RevId: 3f03b8416c95acf2b68da1db56cbe36a513a4bde
## Description
This PR removes `MetadataStorageT`, and cleans up all top-level error handling. In short: this PR changes `MonadMetadataStorage` to explicitly return a bunch of `Either QErr a`, instead of relying on the stack providing a `MonadError QErr`. Since we implement that class on the base monad *below any ExceptT*, this removes a lot of very complicated instances that make assumptions about the shape of the stack.
On the back of this, we can remove several layers of ExceptT from the core of the code, including the one in `RunT`, which allows us to remove several instances of `liftEitherM . runExceptT`.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7689
GitOrigin-RevId: 97d600154d690f58c0b93fb4cc2d30fd383fd8b8
## Description
Adds a content-length response header to all endpoints. This PR tests this feature by checking the content-length of every request we send in the tests.
## Changelog ✍️
__Component__ : server
__Type__: enhancement
__Product__: community-edition
### Short Changelog
add a content-length response header to all endpoints
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7444
Co-authored-by: Manas Agarwal <5352361+manasag@users.noreply.github.com>
GitOrigin-RevId: a0a811852053c5dde4b11b71ba11a7d456c84d76
## Description
This PR updates the JWK refresh thread to poll every second instead of the previous behaviour where the thread used to sleep based on the expiry time in `Cache-Control`/`Expires` response headers.
## Motivation
As a part of dynamically updating environment variables on cloud without restart the user projects, we want to implement a mechanism which makes HGE aware of any changes in the user configuration by updating a shared variable data type which can be accessed by relevant threads/core functionality before their execution.
The above updates requires us to make the threads polling in nature such that before executing their code, any change in the user config is captured and the appropriate behaviour is channelised. In the case of JWK updating thread, the thread used to sleep for the time as mentioned in the `Cache-Control` or `Expires` headers which make the thread unware of any new changes in the user config in that period of time, hence requiring a restart to propogate the new changes.
To solve this problem we have now updated the JWK update thread to poll every second for change in `AuthMode`(from a shared variable in subsequent changes to implement the dynamic env var update feature) and update the JWK accordingly such that it does not use any stale configurations and works without HGE restart.
### Related Issues
https://hasurahq.atlassian.net/browse/GS-300
### Solution and Design
- We store the expiry time in the `JWTCtx`
- On every poll check whether the current time exceeds the expiry time, in which case we call the JWK url to fetch the new JWK and expiry.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7177
Co-authored-by: Krushan Bauva <31391329+krushanbauva@users.noreply.github.com>
Co-authored-by: Anon Ray <616387+ecthiender@users.noreply.github.com>
GitOrigin-RevId: bc1e44a8c3823d7554167a7f01c3ce085646cedb
Hooks up event trigger codecs from #7237. This required fixing a problem where some backend types implemented `defaultTriggerOnReplication` with `error` which caused the server to crash when evaluating those for default values in codecs. The changes here add a type family to `Backend` called `XEventTriggers` that signals backend support for event triggers, and changes the type of `defaultTriggerOnReplication` to from `TriggerOnReplication` to `Maybe (XEventTriggers b, TriggerOnReplication)` so that it can only be implemented with a `Just` value if `XEventTriggers b` is inhabited. This emulates some existing type families in `Backend`. (Thanks to @daniel-chambers for this suggestion!)
I used the implementation of `defaultTriggerOnReplication` as a signal for event triggers support to prune the Metadata API so that event trigger fields will not appear in the OpenAPI spec for backend types that do not support event triggers. The codec version of the API will also not emit or accept those fields for those backend types. I think I could use `Typeable` to test whether `XEventTriggers` is `Void` instead of testing whether `defaultTriggerOnReplication` is `Nothing`. But the codec implementation will crash anyway if `defaultTriggerOnReplication` is `Nothing`.
I checked to make sure that graphql-engine-pro still compiles.
Ticket: https://hasurahq.atlassian.net/browse/GDC-521
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7295
GitOrigin-RevId: 2b2dd44291513266107ca25cf330319bf53a8b66
We currently let the garbage collector and/or the operating system clean up our mess. This is mostly fine in production (kind of) but a problem when we want to start many HGE servers in parallel for testing purposes.
Shutting them down should, in theory, ease the load.
There is more work to be done in the API test suite before this is very helpful. Right now the test suite actually runs the finalizers on the server context straight away and then uses the leaked resources. As there's no way to actually "close" a connection pool, it keeps working regardless. If we wanted to be strict about this we might want to add a "closed" flag to `Data.Pool` which would cause an exception on `withResource` after closing it.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7299
GitOrigin-RevId: ba02f96c7b5b06ba3ba7080a5583a56cb0efcfa7
Codecs for event triggers, including webhook transforms. These are not hooked into the higher-up table metadata codec yet because some backend implementations implement event triggers with `error` which causes an error when codecs are evaluated. I plan to follow up with another PR to resolve that.
Ticket: https://hasurahq.atlassian.net/browse/GDC-585
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7237
GitOrigin-RevId: 8ce40fe6fedcf8b109d6ca50a505333df855a8ce
Generate more Metadata Inconsistencies instead of startup failures. Specifically this means that
- errors retrieving the main query of an executable GraphQL document, and
- errors during fragment inlining
no longer fail irrecoverably.
This also makes more parts of `buildSchemaCacheRule` into pure code, which is always nice.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7234
GitOrigin-RevId: aebf636c2fb1aad1c2df9a37f7d0b67c1ee40c42
context: This is foundation work, before we change how the server chooses to compress or not
part of effort: #5518
-----
Prior to this change it was difficult to understand how the functionality in this module related to the semantics of Accept-Encoding. We also didn't correctly handle directives with qvalues.
After this change certain technical infelicities are called out without modifying the behavior of the server; for instance we continue to fall back to identity (no compression) in the case where technically we're supposed to return 406, and we also continue to treat `*` conservatively as meaning “use no compression”.
The only external change here is `gzip;q=x.y` now results in a zipped response.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7213
GitOrigin-RevId: 1910ffd70d29f1ab8825c601f1bd998be70ceeeb
`toLazyByteString` is a little deficient in two ways:
- It allocates relatively large chunks (4KB + 32KB +32KB, etc…) which is wasteful for small ByteStrings
- It shrinks each chunk (Copying the data to a new chunk of exactly the right size) if it's not more than half filled. If we're running the builder right before we send it over the wire, this copy is totally extraneous (we simply end up with more work for the next GC)
part of the effort: https://github.com/hasura/graphql-engine-mono/issues/5518
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7187
GitOrigin-RevId: b499cd49c33da6cfee96be629a36b5c812486e39
## Description
There is a bug in the metadata defaults code, see [the original PR](https://github.com/hasura/graphql-engine-mono/pull/6286).
Steps to reproduce this issue:
* Start a new HGE project
* Start HGE with a defaults argument: `HASURA_GRAPHQL_LOG_LEVEL=debug cabal run exe:graphql-engine -- serve --enable-console --console-assets-dir=./console/static/dist --metadata-defaults='{"backend_configs": {"dataconnector": {"mongo": {"display_name": "BONGOBB", "uri": "http://localhost:8123"}}}}'`
* Add a source (doesn't need to be related to the defaults)
* Export metadata
* See that the defaults are present in the exported metadata
## Related Issues
* Github Issue: https://github.com/hasura/graphql-engine/issues/9237
* Jira: https://hasurahq.atlassian.net/browse/GDC-647
* Original PR: https://github.com/hasura/graphql-engine-mono/pull/6286
## Solution
* The test for if defaults should be included for metadata api operations has been extended to check for updates
* Metadata inconsistencies have been hidden for `/capabilities` calls on startup
## TODO
* [x] Fix bug
* [x] Write tests
* [x] OSS Metadata Migration to correct persisted data - `server/src-rsr/migrations/47_to_48.sql`
* [x] Cloud Metadata Migration - `pro/server/res/cloud/migrations/6_to_7.sql`
* [x] Bump Catalog Version - `server/src-rsr/catalog_version.txt`
* [x] Update Catalog Versions - `server/src-rsr/catalog_versions.txt` (This will be done by Infra when creating a release)
* [x] Log connection error as it occurs *(Already being logged. Requires `--enabled-log-types startup,webhook-log,websocket-log,http-log,data-connector-log`)
* [x] Don't mark metadata inconsistencies for this call.
## Questions
* [ ] Does the `pro/server/res/cloud/migrations/6_to_7.sql` cover the cloud scenarios?
* [ ] Should we have `SET search_path` in migrations?
* [x] What should be in `server/src-rsr/catalog_versions.txt`?
## Testing
To test the solution locally run:
> docker compose up -d
and
> cabal run -- exe:api-tests --skip BigQuery --skip SQLServer --skip '/Test.API.Explain/Postgres/'
## Solution
In `runMetadataQuery` in `server/src-lib/Hasura/Server/API/Metadata.hs`:
```diff
- if (exportsMetadata _rqlMetadata)
+ if (exportsMetadata _rqlMetadata || queryModifiesMetadata _rqlMetadata)
```
This ensures that defaults aren't present in operations that serialise metadata.
Note: You might think that `X_add_source` would need the defaults to be present to add a source that references the defaults, but since the resolution occurs in the schema-cache building phase, the defaults can be excluded for the metadata modifications required for `X_add_source`.
In addition to the code-change, a metadata migration has been introduced in order to clean up serialised defaults.
The following scenarios need to be considered for both OSS and Cloud:
* The user has not had defaults serialised
* The user has had the defaults serialised and no other backends configured
* The user has had the defaults serialised and has also configured other backends
We want to remove as much of the metadata as possible without any user-specified data and this should be reflected in migration `server/src-rsr/migrations/47_to_48.sql`.
## Server checklist
### Catalog upgrade
Does this PR change Hasura Catalog version?
- ✅ Yes
### Metadata
Does this PR add a new Metadata feature?
- ✅ No
### GraphQL
- ✅ No new GraphQL schema is generated
### Breaking changes
- ✅ No Breaking changes
## Changelog
__Component__ : server
__Type__: bugfix
__Product__: community-edition
### Short Changelog
Fixes a metadata defaults serialization bug and introduces a metadata migration to correct data that has been persisted due to the bug.
PR-URL: https://github.com/hasura/graphql-engine-mono/pull/7034
GitOrigin-RevId: ad7d4f748397a1a607f2c0c886bf0fbbc3f873f2
