There are several maintenance scripts waiting to be run on many wikis since years ago: https://bugzilla.wikimedia.org/showdependencytree.cgi?id=29782&hide_reso... The biggest offenders are: * https://bugzilla.wikimedia.org/show_bug.cgi?id=16112 which breaks tons of special pages; * https://bugzilla.wikimedia.org/show_bug.cgi?id=33253 which makes the article count on those wikis completely broken after the count method change.
The fact that we could/should find a perfect and permanent solution for those problems doesn't mean that we shouldn't improve the situation given that we can. The fact that we can't run them every month as we once did doesn't mean that we shouldn't run them say every 5 years (which would be an improvement) or after big changes (such as the count method). Even the longest scripts of that sort, when run by pdhanda some months ago, took only some days for all wikis and didn't do any damage.
Nemo
If the script is optimized per Brion recommendations, I think we could execute it immediately... I don't really know why ops didn't do that, I was in though that it's cronned to run certain jobs
On Fri, Jan 27, 2012 at 11:45 AM, Federico Leva (Nemo) nemowiki@gmail.comwrote:
There are several maintenance scripts waiting to be run on many wikis since years ago: https://bugzilla.wikimedia.** org/showdependencytree.cgi?id=**29782&hide_resolved=1https://bugzilla.wikimedia.org/showdependencytree.cgi?id=29782&hide_resolved=1 The biggest offenders are:
- https://bugzilla.wikimedia.**org/show_bug.cgi?id=16112https://bugzilla.wikimedia.org/show_bug.cgi?id=16112which breaks tons of special pages;
- https://bugzilla.wikimedia.**org/show_bug.cgi?id=33253https://bugzilla.wikimedia.org/show_bug.cgi?id=33253which makes the article count on those wikis completely broken after the
count method change.
The fact that we could/should find a perfect and permanent solution for those problems doesn't mean that we shouldn't improve the situation given that we can. The fact that we can't run them every month as we once did doesn't mean that we shouldn't run them say every 5 years (which would be an improvement) or after big changes (such as the count method). Even the longest scripts of that sort, when run by pdhanda some months ago, took only some days for all wikis and didn't do any damage.
Nemo
______________________________**_________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/**mailman/listinfo/wikitech-lhttps://lists.wikimedia.org/mailman/listinfo/wikitech-l
On Fri, Jan 27, 2012 at 8:57 PM, Petr Bena benapetr@gmail.com wrote:
execute it immediately... I don't really know why ops didn't do that, I was in though that it's cronned to run certain jobs
It is cronned for only for the small wikis, I believe[Citation Needed].
I don't think it's ever executed on small wikis either. We have 5 years old ghost entries in Wanted* special pages on it.quote. They're like family friends now, but still ghosts.
Nemo
I believe that as soon as people from ops manage to move configuration to gerrit and puppet we would have more opportunity to improve the current configuration at least on labs where even crontab would be replicated, and properly optimize it On Jan 28, 2012 3:01 PM, "Federico Leva (Nemo)" nemowiki@gmail.com wrote:
I don't think it's ever executed on small wikis either. We have 5 years old ghost entries in Wanted* special pages on it.quote. They're like family friends now, but still ghosts.
Nemo
______________________________**_________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/**mailman/listinfo/wikitech-lhttps://lists.wikimedia.org/mailman/listinfo/wikitech-l
wikitech-l@lists.wikimedia.org