release-21.1: stop: break deadlock between Stopper.mu and Replica.mu #63871
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Outdated suggestions cannot be applied.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.
Backport 1/1 commits from #63867.
/cc @cockroachdb/release
Fixes #63761. As of #61279, it was possible for us to deadlock due to
inconsistent lock orderings between Stopper.mu and Replica.mu. We were
previously holding onto Stopper.mu while executing all closers,
including those that may acquire other locks. Because closers can be
defined anywhere (and may consequently grab any in-scope lock), we
should order Stopper.mu to come after all other locks in the system.
The closer added in #61279 iterated over all non-destroyed replicas,
locking Replica.mu to check for the replica's destroy status (in
Store.VisitReplicas). This deadlocked with the lease acquisition code
path that first grabs an exclusive lock over Replica.mu (see
InitOrJoinRequest), and uses the stopper to kick off an async task
acquiring the lease. The stopper internally locks Stopper.mu to check
whether or not it was already stopped.
Release note: None