See P1968 for log. CN message groups alone about two hours to index, which is more than all other message groups across all sites.
There seems to be something blocking there, because even though I ran with 4 threads, only one of them seemed to be doing something at a time during CN groups. You can see this in the log by having bunch of groups complete (not shown in the log) at the same time, and new threads spawned for new groups at the same time (shown in the log).
| Status | Assigned | Task | ||
|---|---|---|---|---|
| Open | None | T34482 CentralNotice should respect language fallbacks | ||
| Open | None | T38591 Use the Translate extension in CentralNotice (tracking) | ||
| Open | None | T111189 Central Notice message groups are slow to index | ||
| Open | None | T116235 [Epic] CentralNotice translation should move closer to MediaWiki i18n standards and the code cleaned up |
You added me to CC so I assume this is High priority at least?
/me is just wondering why he was CC'd :)
Hi Greg, I was planning to send you an email but forgot. To put it short, I had some delay in beginning due to the script refusing to run. After that all went fine except reindexing meta took more than two hours so I went a quite bit of overtime of the allocated window. This issue is not urgent, but if not looked into I might require to ask 4h slot next time ;)
Hi @Nikerabbit, thanks for noticing this! We're happy to support your investigation in any way we can, it sounds like there's something wrong with the way we're using message groups, perhaps we're spamming or there's a problem with the index schema? Please let us know whatever you find and if we can do anything.