Opened on 10/19/2015 at 01:39:57 PM
Closed on 10/28/2015 at 06:35:09 PM
#3211 closed change (fixed)
ensure filterlist servers deliver notification.json
Reported by: | sporz | Assignee: | matze |
---|---|---|---|
Priority: | P2 | Milestone: | |
Module: | Infrastructure | Keywords: | |
Cc: | fred, kirill, fhd, trev | Blocked By: | |
Blocking: | #2955, #3115 | Platform: | Unknown / Cross platform |
Ready: | yes | Confidential: | no |
Tester: | Unknown | Verified working: | no |
Review URL(s): |
Description (last modified by matze)
The FCGI daemon on some filter servers died silently which resulted in clients not being able to download notification.json. The other resources like exceptionrules.txt were not affected, which resulted in biased statistics for the data module. See also #3209.
Todo
Setup a mechanism to ensure crashed daemons being restarted and integrate it with the notification handler.
Attachments (0)
Change History (6)
comment:1 Changed on 10/19/2015 at 01:50:36 PM by matze
- Cc fred added; matze removed
- Owner set to matze
- Priority changed from Unknown to P2
- Ready set
comment:2 Changed on 10/19/2015 at 02:14:58 PM by matze
- Blocking 2955, 3115 added
- Cc fhd added
- Description modified (diff)
comment:3 Changed on 10/19/2015 at 03:00:19 PM by trev
- Cc trev added
We already have sitescripts.management.bin.start_services which is running on server16. It is being configured via sitescripts.ini.
comment:4 Changed on 10/19/2015 at 03:06:39 PM by matze
Thank you, @felix mentioned that earlier today already, it will become the basis of this patch.
comment:5 Changed on 10/20/2015 at 10:51:12 PM by matze
- Review URL(s) modified (diff)
- Status changed from new to reviewing
comment:6 Changed on 10/28/2015 at 06:35:09 PM by matze
- Resolution set to fixed
- Status changed from reviewing to closed
The same should be done with other services, e.g. hgweb, rietveld and trac. Thus there is definitely a need for a solution more sophisticated than hard-coding some script or cron-job per case.