We should probably get some input from SRE on this one.

I think we should keep it !log'ing. We log all MediaWiki code deploys through scap, and l10nupdate shouldn't be an exception.

Ok, repurposing this task for whether or not we should remove it from !log'ing.

I still think alerting (more than just a !log) is needed for the l10nupdate, though, since it's automatic and happens when most people are not online.

I'd like to continue getting l10nupdate updates on the !log lines that flow to the #wikimedia-operations channel (or equivalent) if we can. It's good for correlating with any related issues, and I suspect even now l10update timing may be related to some difficult-to-track down low-impact issues we've been investigating for a while with timeout spikes from mw->varnish for Main_Page, etc...

In T107480#1496901, @Legoktm wrote:

I think we should keep it !log'ing. We log all MediaWiki code deploys through scap, and l10nupdate shouldn't be an exception.

+1
Given LocalisationUpdate failures are usually noticed by translators rather than sysadmins/devs/whatever, SAL is the most reasonable place.

As for the noise, this proposal appears to be a (non|wrong) solution for T49301.