You should provide exact configuration settings related to cache and localization caching.

Also, clarify what do you mean by many times if the cache is not rebuilt there are LC cache issues... when does that happen? when you upgrade?

Hi @Ciencia_Al_Poder and @Aklapper,

I'm NDKilla, one of Orain.org's System Administrators (at least, on our mediawiki servers. I don't have global root access).

This has definitely been annoying for us and I was actually just made aware of this issue. I just read the link Aklapper provided and will try to clarify some things..

I'm not exactly sure how or why this happens, but the results are easy enough to occasionally view. If you head over to http://orain.org and login, clicking the Extension:Echo notification window will display broken messages like in this image

This is also noticeable in a few places on our wiki (anywhere custom messages are used) like Special:Logs/farmer

Orain's Ansible Playbook automatically updates (and as needed, downloads new) extensions every hour on the hour, and this may be part of the problem. The ansible playbook is also supposed to automatically rebuild the cache, but it doesn't force rebuild any languages that "don't need to be rebuilt" according to the software, and this appears to be a problem.

Frequently (several times a day, but usually not every hour) the LC cache appears to degrade. running the rebuildLocalisationCache.php maintenance script on our wiki that has all extensions enabled appears to fix the issue globally, but only temporarily.

I'm not sure what cache settings you'd like to see, but many of our settings can be found in our ansible playbook's mediawiki files online https://github.com/Orain/ansible-playbook/tree/master/roles/mediawiki/files/localsettings

I recently added a cron job to auto force a refresh of the LC caches on all MW servers, this has minimized the disruption caused by this issue, but it's merely a stopgap at best, it has done nothing to fix the underlying problem.

@Arcane21 I'm fairly certain that cron job was removed ~10 days ago by addshore
I've manually been running the LC rebuild on extloadwiki 2-3 times a day for a week every day I'm home

In T110345#1591268, @NDKilla wrote:

@Arcane21 I'm fairly certain that cron job was removed ~10 days ago by addshore
I've manually been running the LC rebuild on extloadwiki 2-3 times a day for a week every day I'm home

That would explain why I've been seeing the system messages again more often.

I'm curious if the conversation in https://github.com/Orain/ansible-playbook/issues/650 is related?

@Aklapper definitely related. More or less when it started

@Aklapper I've replied to a question on that github issue related to our config settings, might be interested

https://github.com/Orain/ansible-playbook/pull/720 hopefully fixed it
unlike github, should wait here before closing this IMO, but I feel rather silly for this :p