Unsafe data race caused by the global TSO request pool #8076
Labels
affects-8.1
severity/critical
The issue's severity is critical.
type/bug
The issue is confirmed as a bug.
Projects
Bug Report
If
(*client).dispatchTSORequestWithRetry
directly callstryDone()
due to the error of context canceled, the caller might get the TSO result and put the request back into the pool. At this time, the oldtsoClient
may have data races with the newtsoClient
when cleaning up residual TSO requests.This is also the root cause of #8055.
The text was updated successfully, but these errors were encountered: