Something is broken with either SMW or cleanup of instance wiki pages.
Description
Details
Status | Subtype | Assigned | Task | ||
---|---|---|---|---|---|
Resolved | Andrew | T139773 Labs front-page statistics are very wrong | |||
Resolved | Andrew | T140298 Actually delete instance status pages when the instance is destroyed | |||
Resolved | PRODUCTION ERROR | Nikerabbit | T95839 CAS update failed on user_touched for user ID | ||
Resolved | PRODUCTION ERROR | None | T150373 CAS update failed on user_touched (triggered from ApiLogin on Wikitech) |
Event Timeline
I guess I know at least one reason: If you delete a instance at horizon, the wikitech won't get deleted, labslogbot only adds sometimes the "instance has been deleted" note. Guess we need kind of synch.
At least some of these are caused by a race in page maintenance. I can see cases where the page is cleared and the instance marked deleted, and right after the page is refreshed by the periodic update job.
@ If you delete a instance at horizon, the wikitech won't get deleted
That is incorrect -- the pages are deleted by a callback within nova that's triggered by deletion. The callback is bound to the DNS entry cleanup code, so I suspect we leak a page at the same time that we leak a DNS entry.
Change 321677 had a related patch set uploaded (by Andrew Bogott):
wikistatus: Replace the max_retries throttle for wiki login
Change 321677 merged by Andrew Bogott:
wikistatus: Replace the max_retries throttle for wiki login
Change 322208 had a related patch set uploaded (by Andrew Bogott):
wikistatus: Further attempt to solve login races
Change 322208 merged by Andrew Bogott:
wikistatus: Further attempt to solve login races