Deadlock in LM by using nbackup utility [CORE1696] #2121
Assignees
Comments
|
Commented by: @romansimakov To resolve the problem nbackup must have an order of lock requesting for LCK_bdb and LCK_backup_database. |
Modified by: @pcisarWorkflow: jira [ 13831 ] => Firebird [ 14128 ] |
Modified by: @romansimakovtimeestimate: 86400 [ 86400 ] => timeoriginalestimate: 86400 [ 86400 ] => |
Modified by: @romansimakovassignee: Roman Simakov [ roman-simakov ] |
Modified by: @romansimakovstatus: Open [ 1 ] => Resolved [ 5 ] resolution: Fixed [ 1 ] Fix Version: 2.5 Beta 1 [ 10251 ] |
Modified by: @pcisarstatus: Resolved [ 5 ] => Closed [ 6 ] |
Modified by: @pavel-zotovQA Status: No test |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Submitted by: @romansimakov
Deadlock description: Process A inserts records into a table. It keeps LCK_read lock on LCK_backup_database and can fetch other pages since NBackupState use recursive locks. So process A calls FETCH_LOCK for a page (for example PP) and waits it if a process B already uses this page. On the other hand process B can have called FETCH_LOCK and must read the page from disk. To do this it must request LCK_backup_database with LCK_read level for status receiving. In this case process B have to wait if third process N already request LCK_backup_database with LCK_write level.
Commits: 1a49498
The text was updated successfully, but these errors were encountered: