Forum:Help:Job queue

i've been reading the help topic on job queue's and there's one thing i need to know

each time a request runs, one job is taken from the job queue and executed

What does it mean by request?

-- Semaj draehs - Talk page 11:18, 19 August 2008 (UTC)
 * a page load. the whole queue is also run about every 24 hours or so, sometimes longer between. --Uberfuzzy 13:44, 19 August 2008 (UTC)

Is there anyway to speed it up as ours is currently at 1003...

-- Semaj draehs - Talk page 14:05, 19 August 2008 (UTC)


 * One time we had more than 5 times the numbers of page in fr.guildwars in the queue. — TulipVorlax 17:20, 19 August 2008 (UTC)


 * Put the whole phrase here: By default, each time a request runs, one job is taken from the job queue and executed. That's not the case of Wikia. In fact, seems like the $wgJobRunRate variable is set to 0, so the job queue never is processed on a request, because edits doesn't seem to reduce it. Also, the server cache servers prevents pageviews to hit the Apache servers, so they doesn't touch the job queue. It seems like the job queue is processed by a cron job every day, or so. --Ciencia Al Poder (talk) -WikiDex 18:32, 19 August 2008 (UTC)

Thats what i found suspicous, when i heard page load because i edited the wiki a lot and the job que stayed the same, is there a way to heighten the job que so it does run at 1 job executed at every page edit?

-- Semaj draehs - Talk page 19:33, 19 August 2008 (UTC)


 * For simple users ou simple sysop, no there's no way we can alter how the software is functionning. But you still ask a staff to see what he will awnser about the matter. — TulipVorlax 03:19, 21 August 2008 (UTC)

What happened in the past few days to cause the job queue to be emptied?
A day or two ago I noticed the job queue at rappel.wikia.com was at zero. That is the first time I'd seen it reach zero since the mediawiki software upgrade a couple of months ago. It had been steadily increasing in the 8000-9000 range

I have read the mediawiki help and help_talk page discussions about the job queue. I am not sure that I have a good understanding of my wiki's job queue. If you can point me to a layman's explanation then it may explain why I sometimes have to open a page for edit and save it without making any change just to see the effect of a recently edited template. The talk page discussion link above is the easiest for me to understand but since it is posed as a question and not as an authoritative response I am left guessing if this is one of those cases of the blind leading the blind.

If a site administrator can cause (or request) the job-queue to be processed and emptied then I'd like to know how to do that (or where to submit a request that this be done).

The rappelz wiki makes extensive use of templates (about 350 of them) to centrally store data about each in-game skill. It makes use of 3 templates to present that data in one of three uniform formats (2/3 are table formats) on at least 1 page per skill plus up to 27 job pages. Therefore, when a specific template is changed that change might propagate to anywhere between 1 and ~380 pages.

If I understood the job queue better then I may be able to modify my template editing practices to lessen the server load wrt these edits. Alternatively I may be able to routinely empty the job queue at some pre agreed time of day when the server load is not very high.


 * The mediawiki description is simply not consistent with behavior observed here at wikia.

Also, in the past two weeks I have seen the error message: "Wikia is sadly having problems. Our backend servers are not responding. We are working to fix it as soon as possible." on about 8/10 days that I have been here. Is the almost daily frequency of these errors related to the job queue for all wikis?


 * 1) Could somebody who understands the mechanics of how the job queue operates here at wikia please provide a layman's description that improves upon the mediawiki description?
 * 2) Could a wikia staff member please advise the frequency that the job queue is emptied completely.

najevi 04:45, 3 November 2008 (UTC)
 * No, the 'backend' problem are not related. That is a different hardware related issue that is being resolved.
 * No, there is no way for admins to manually run the queue, it is handled by a automated timing system.
 * I think, last time i checked, it ran every 48 hours (approx). But there are some other scaling issues with the job queue system that are being rewritten to make things run more smoothly for everyone involved. You are not alone in this problem, wikis larger and smaller are being affected by this.
 * The mediawiki description of the system is for the system that handles the job queue for 1 wiki (wikipedia). Wikia's setup is slightly different. Like i said, its being further modified to help it scale even farther then it is now.
 * Ideally, admins should not have to worry about about backend issues, especially ones as low level as the job queue. I applaud you for taking the extra steps in reading up about it and considering it, but theres nothing that can be done at this level.
 * --Uberfuzzy 05:45, 3 November 2008 (UTC)