This is (yet another) side effect of wikitech being so special. The updating process only runs in the currently active datacenter (as-of now, codfw), while wikitech currently only lives in eqiad (there's no codfw equivalent of wikitech as of writing). That makes the updating process for wikitech error up with "DB is read only", as it runs in codfw (as that's where most wikis are). The updating process will run again normally once we're back in eqiad (per T287539, will happen on Sept 14, ie in two days).

T167973: Move database for wikitech (labswiki) to a main cluster section would be a real solution to this issue (apart from special-casing wikitech in the config, letting it run always in eqiad, regardless of what is the active DC). It appears that task will be worked on shortly after we're switched back. As such, I'm boldly declining this task – before the next update takes place, it will be automatically fixed at least by the switchback.

If you want, I can do a manual update tomorrow – otherwise, waiting for the switchback (and the update process to kick in).