Fix connection race in concurrent GetLastTimeStamp#387
Merged
SimonCropp merged 2 commits intomainfrom Apr 21, 2026
Merged
Conversation
Commit d9016cc cached ResolveQuery's Task via Interlocked.CompareExchange so concurrent callers would share a single query. But each caller that lost the CompareExchange race still had its own ResolveQuery task running in the background, executing Open/HasViewServerState/Close against the caller's own connection. Meanwhile the caller proceeded to Execute against that same connection, producing "The connection is closed" (production) or "does not support MultipleActiveResultSets" (local) errors. Serialize ResolveQuery with a SemaphoreSlim so at most one caller ever runs it. Strengthen ConcurrencyTests to use closed connections, a synchronous barrier on dedicated threads, and 20 iterations — reliably reproduces the race on the old code.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Commit d9016cc cached ResolveQuery's Task via Interlocked.CompareExchange
so concurrent callers would share a single query. But each caller that
lost the CompareExchange race still had its own ResolveQuery task running
in the background, executing Open/HasViewServerState/Close against the
caller's own connection. Meanwhile the caller proceeded to Execute against
that same connection, producing "The connection is closed" (production)
or "does not support MultipleActiveResultSets" (local) errors.
Serialize ResolveQuery with a SemaphoreSlim so at most one caller ever
runs it. Strengthen ConcurrencyTests to use closed connections, a
synchronous barrier on dedicated threads, and 20 iterations — reliably
reproduces the race on the old code.