Make all store/db interactions async #6194
Open
+7,210
−5,463
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
lutter
force-pushed
the
lutter/async-store
branch
6 times, most recently
from
2be39ef to
63352af
Compare
lutter
force-pushed
the
lutter/async-store
branch
5 times, most recently
from
5877de5 to
212b69c
Compare
Instead of reexporting from the graph crate, declare tokio once in the workspace and use it as needed
We need to make sure that all tests usea single runtime. See the comment on TEST_RUNTIME for why that's needed
That ensures that all tests use the same runtime, and we therefore don't have issues with closed connections because a second runtime was dropped.
The test assumed that the chain is in the primary, which is not necessarily true when tests are run against a sharded setup
lutter
force-pushed
the
lutter/async-store
branch
from
212b69c to
9f0ce51
Compare
deadpool seems to get much more use and is udated more frequently. It also provides better insights on various size etc. metrics
For now, this is just a copy of AsyncDieselConnectionManager
lutter
force-pushed
the
lutter/async-store
branch
from
321656a to
9356607
Compare
lutter
force-pushed
the
lutter/async-store
branch
from
9356607 to
6e768dc
Compare
lutter
changed the title
We have runner tests failing frequently because the store seems to get dropped while it still has pending writes. This change attempts to prevent that by flushing the store when we encounter a fatal error.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
We have a lot of places where we do blocking database work from a non-blocking tokio task. This PR addresses this longstanding issue by using async database connections and
diesel_async. With that, we don't need to rely on error-prone approaches that require spawning blocking tasks whenever we do database work.Most of this PR is concerned with sprinkling async/await throughout the code base. There are a lot of commits, but most of them are boring, and I tried to structure it so that many of them just add async/await in a few places to make them easy to review individually.
Since switching to async connections also requires changing the connection pool, towards the end of the PR there are changes to recreate the instrumentation we had with r2d2 with deadpool.
There are still places in the codebase where
spawn_blockingis used, but those can be removed in separate PR's as this one is already crazy long. My apologies to the reviewer(s)I made sure that
cargo checkpassed after each commit; in addition, I ran CI after every few (say 5-10) commits and made sure that passed, too, up to the commit markedSQUASH FROM HEREas it was not possible to pass CI from that point on untilCI PASSES AGAINas theAsyncConnectionWrapperusesblock_onfor db interactions, and too many of them happen on the main tokio threads. For those, tokio panics as you are not allowed to useblock_onon those threads.Fixes #905