-
Bug
-
Resolution: Fixed
-
Low
-
3.11.4
Problem
When deleting a user directory the entire instance will lock up, all requests will wait forever.
This is caused by a database deadlock in crowd which blocks all calls to things that do things like find the current authenticated user.
Affected Environments
Evaluation users using HSQL or H2 where table level locking is used.
Cannot reproduce on 4.2-rc7 So changes in the crowd code since 3.11 may have reduced or eliminated the chances of this error occurring.
Workaround
- Restart the instance.
- Upgrade to one of the supported production databases. H2 and HSQLDB are for evaluation purposes only.
Form Name |
---|
If any customer sees this in a version after 4.2.0, please let us know.
Please note however that both H2 and HSQL are both unsupported databases. You must upgrade to a supported database for production deployments of Bitbucket Server.
Further, we are unable to reproduce this issue in 4.2+ hence we are closing this issue.