Swarm manager should time out idle agent connections
Nobody has claimed this yet.
- Dominant language
- Go
- Stars
- 3.7k
- Forks
- 676
- Avg merge
- 4d 9h
- Merged PRs (30d)
- 6
Description
The manager should time out old connections that have not have recent activity. One possible source of these connections is a bug like #1274. Another example of where this could happen is an agent with a flaky network that isn't able to hold open a connection, and keeps opening new ones in retry attempts.
Since agents send a periodic heartbeat, connections with no recent activity should be considered stale and they should be cleaned up.
In the current state of affairs, these connections live on forever unless something happens at the TCP level to terminate them, and the manager will eventually have its resources exhausted.
I'm not sure whether GRPC supports this, but if it doesn't we should request it. It's an important feature.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No files, tests, or entry points are named. Begin by locating the swarm manager's agent-connection handling and checking whether gRPC provides an idle-connection timeout. Done means stale connections are cleaned up based on heartbeat inactivity; if gRPC lacks support, the issue should lead to a support request.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- go, grpc
- Domain
- backend, distributed-systems
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 25/100