Title | Bugzilla integration holds locks after errors |
Status | closed |
Priority | essential |
Assigned user | Nick Barnes |
Organization | Ravenbrook |
Description | The Bugzilla integration locks some Bugzilla tables during each replicator poll. If there is an error during the poll, the tables will remain locked until the next poll begins. This may be a long time if the replicator is failing and backing off exponentially. While the tables are locked, Bugzilla is unusable. |
Analysis | This is Perforce job 6261. Need some sort of "poll failure" method in the defect tracker interface, called by the replicator if there's any sort of replicator failure during a poll cycle. |
How found | manual_test |
Evidence | Our test server got bogged down and Bugzilla stopped working. I couldn't reproduce this, and MySQL doesn't seem to have any sort of lock debugging or diagnostic tools, but it definitely seemed to be a locking failure. |
Observed in | 1.0.0 |
Introduced in | 0.5.0 |
Created by | Nick Barnes |
Created on | 2001-04-23 13:22:10 |
Last modified by | Gareth Rees |
Last modified on | 2001-12-10 19:40:28 |
History | 2001-04-23 NB Created. 2001-07-17 GDR Downgraded to "essential" since people can get by. |
Change | Effect | Date | User | Description |
---|---|---|---|---|
23709 | closed | 2001-11-01 14:09:08 | Nick Barnes | Add poll_start and poll_end to the defect tracker interface. |