Opened on 08/26/2015 at 03:28:40 PM
Closed on 03/18/2018 at 02:32:50 PM
#2955 closed defect (incomplete)
Rietveld Resource Leak
Reported by: | matze | Assignee: | |
---|---|---|---|
Priority: | P4 | Milestone: | |
Module: | Infrastructure | Keywords: | |
Cc: | fhd, fred, saroyanm | Blocked By: | #3211 |
Blocking: | Platform: | Unknown / Cross platform | |
Ready: | yes | Confidential: | no |
Tester: | Unknown | Verified working: | no |
Review URL(s): |
Description (last modified by matze)
Our Rietveld setup seems to suffer from a scaling issue. Obviously just a few dozen concurrent requests can already cause the system to zombificate any of the appserver workers, which results in both memory and I/O overload quite fast.
Attachments (0)
Change History (4)
Note: See
TracTickets for help on using
tickets.
Manual restarting (including a delay for the I/O to recover, so no direct restart) solved the issue for now. Thus I do not consider it very critical right now. A daily restart via cron would probably hide the symptoms for a few years, given the current usage growth.
Anyway, we should of course tackle the actual issue itself. Whatever that may be.