Temporary memory leak in FutureImpl.executors

[jglick]

Alarmed by this thread I tried to see if I could reproduce a memory leak from running K8s builds with WebSocket. I could not reproduce any major leak, but did find some minor issues.

Testing done

Without patch, the test fails with this reference chain:

private static jenkins.model.Jenkins jenkins.model.Jenkins.theInstance->
hudson.model.Hudson@39f8cf44-queue->
hudson.model.Queue@d78af81-leftItems->
com.google.common.cache.LocalCache$LocalManualCache@7a55a792-localCache->
com.google.common.cache.LocalCache@4816e261-segments->
[Lcom.google.common.cache.LocalCache$Segment;@de1aeaf-[1]->
com.google.common.cache.LocalCache$Segment@619f68f4-table->
java.util.concurrent.atomic.AtomicReferenceArray@5d30ba49-array->
[Ljava.lang.Object;@251e4e2e-[2]->
com.google.common.cache.LocalCache$StrongWriteEntry@e03c78d-valueReference->
com.google.common.cache.LocalCache$StrongValueReference@586c2859-referent->
hudson.model.Queue$LeftItem@6425b839-future->
hudson.model.queue.FutureImpl@7665f1fd-executors->
java.util.HashSet@76f2bac1-map->
java.util.HashMap@1ebad322-table->
[Ljava.util.HashMap$Node;@49eb0aac-[2]->
java.util.HashMap$Node@5f7e9273-key->
hudson.model.Executor@94d52cf-owner->
hudson.slaves.SlaveComputer@65397da8

LeftItems are kept around for 5m.

Proposed changelog entries

• Some agent-related objects could be kept in memory after being disconnected and removed from the computer list.

Before the changes are marked as ready-for-merge:

Maintainer checklist

Beta Give feedback

1. There are at least two (2) approvals for the pull request and no outstanding requests for change.
2. Conversations in the pull request are over, or it is explicit that a reviewer is not blocking the change.
3. Changelog entries in the pull request title and/or Proposed changelog entries are accurate, human-readable, and in the imperative mood.
4. Proper changelog labels are set so that the changelog can be generated automatically.
5. If the change needs additional upgrade steps from users, the upgrade-guide-needed label is set and there is a Proposed upgrade guidelines section in the pull request title (see example).
6. If it would make sense to backport the change to LTS, a Jira issue must exist, be a Bug or Improvement, and be labeled as lts-candidate to be considered (see query).

[jglick]

Just in case there was some problem with the Channel separately from the Computer. I could not reproduce any in this test, other than jenkinsci/remoting#694 which was reproduced interactively.

[basil]

jenkinsci/jenkins-test-harness#643 (comment)

[jglick]

Yes jenkinsci/jenkins-test-harness#672 is still on my list of things to look at, but I believe it is only tangentially related to this PR in that they both happen to involve code run from WorkUnitContext.synchronizeEnd; this PR is about the executors list, which is used only if Future.cancel is called, and does not interact directly with the Future done condition.

[timja]

/label ready-for-merge

---

This PR is now ready for merge, after ~24 hours, we will merge it if there's no negative feedback.

Thanks!