Add gRPC connect timeouts, prevent global locking of job_state - #1115
Closed
cmackenzie1 wants to merge 5 commits into
Closed
Add gRPC connect timeouts, prevent global locking of job_state#1115cmackenzie1 wants to merge 5 commits into
cmackenzie1 wants to merge 5 commits into
Conversation
Recovered jobs could stop immediately after logging that their state machine had started. A stale worker address left tonic's connection future pending forever because it had no connection deadline. Apply a 10-second timeout to each attempt while retaining the existing retry and recovery behavior. Exhausted attempts now let the state machine handle the unavailable leader instead of remaining stuck before its first state executes.
A full state-machine queue caused controller RPC handlers to wait while holding the global job map mutex. One stalled job could therefore prevent messages from reaching every other job. Clone the selected job's sender under the map lock, then release the lock before waiting for channel capacity. This preserves per-job backpressure without making it controller-wide.
The recovered-leader fix bounded only one connection path, while other controller and worker clients could still wait indefinitely for tonic to establish a transport connection. Add a global `grpc.connect-timeout` setting with a 10-second default and apply it to every endpoint produced by `grpc_channel_builder`. Remove the leader-specific wrapper while retaining its retry policy, and cover both the default and environment override configuration.
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
mwylde
requested changes
Jul 28, 2026
The global gRPC connection timeout affected every Arroyo service and RPC client even though the observed stall occurs while reconnecting a recovered job to its worker leader. Remove the global configuration and endpoint behavior. Restore the 10-second deadline around each leader connection attempt while preserving the existing retry and recovery behavior.
mwylde
reviewed
Jul 28, 2026
A mandatory global timeout changed connection behavior for every service, while a leader-only fallback prevented deployments from choosing tonic's unbounded default. Add an optional global `grpc.connect-timeout` setting that defaults to unset and apply it when building endpoints. Remove the leader-specific fallback so all clients consistently follow the configured policy.
This was referenced Jul 29, 2026
Collaborator
Author
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Uh oh!
There was an error while loading. Please reload this page.