-
Notifications
You must be signed in to change notification settings - Fork 169
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Data race on transaction commit after sync client shutdown #6844
Comments
@michael-wb @jbreams does this look wrong? In #6293 shouldn't access to SessionWrapper be guarded? |
Yeah, I think the only thing that's really safe to do from outside the event loop is put a bind_ptr to the SessionWrapper in the list of sessions to finalize or actualize. |
It actually also happens just during local run of object store tests when "no_sessions" assertion is hit from #6271 (which was presumably fixed but is still happening). Discovered here #6911 with new added tsan builders for macos. That is what i see also on local macos setup. Is this really a serious issue? Could this lead to some serious problems with data consistency in the process? |
one more situation when this happens on force_close although also through commit_write:
|
One more possible scenario where this could be an issue: SessionWrapper::on_commit and ClientImpl::actualize_and_finalize_session_wrappers
|
Expected results
No tsan warning
Actual Results
Thread Sanitizer reports data race like this:
Steps & Code to Reproduce
Happened a few times during object store tests run. Didn't manage to reproduce this, but probably related to network connection loss during testing. All test cases still pass.
Core version
Core version: 13.17.0
The text was updated successfully, but these errors were encountered: