Erreurs 502 lors du renommage de grandes catégories (par nombre de sujets)

[SimeonGriggs]

Thanks … but these tail'd logs are going a million mile an hour, I don’t think I’ll catch anything

I just tried it again, with the /logs/ open, and there’s no new errors, just new warnings. Mostly:

• “CSP Violation” in a number of different topics.
• “Uncaught [object Object]” in ember_jquery

[pfaffman (Jay Pfaffman)]

Perhaps search the logs for the category name?

[lucasbasquerotto]

@SimeonGriggs

Just trying to rename one category gives a 502 error every time.

This happens with any category or one specific category?

If you create a new category and rename it the error happens?

[Canapin (Coin-coin le Canapin)]

@SimeonGriggs

Just trying to rename one category gives a 502 error every time.

Do your categories have a lot of posts?

[SimeonGriggs]

@lucasbasquerotto

This happens with any category or one specific category?

If you create a new category and rename it the error happens?

Just tried renaming a smaller Category and it worked fine.

@Canapin

Do your categories have a lot of posts?

The one I’m trying to rename sure did.

@pfaffman

Perhaps search the logs for the category name?

Nothing there

[Canapin (Coin-coin le Canapin)]

@SimeonGriggs

The one I’m trying to rename sure did.

Same issue as me then I guess:

502 errors when renaming large (by topic count) categories

Just a correction about this: I think I previously did Category.where(id: 21).update(name: "General") for a category from which I moved the topics out or something like that, because I had to use this command multiple times on various categories right now, and it appears that the command duration is proportional to the number of posts in the target category. So basically, it’s the same issue as when I update the category from the interface, except that it won’t timeout in the command line. It…

[riking (Kane York)]

Summary:

Something during category renames is touching every topic or every post and causing the operation to time out.

[Falco]

It looks related to the search stuff? That will indeed be very costly and should be moved into a background task.

[SimeonGriggs]

Hope not, I fear there are already too many of those.

My server-side backups keep failing because even though they’re running at 2am (local time for the majority of users) there’s too much disk activity for it to complete.

…but I don’t want to get this off topic

[sam (Sam Saffron)]

It is this:

@tgxworld can you kick all the search work including setting version to 0 to a background job?

[tgxworld (Alan Tan)]

PR up at https://github.com/discourse/discourse/pull/11161.

EDIT: Merged

[SimeonGriggs]

This appears to be fixed in the latest update. Thanks team!