this post was submitted on 18 Jul 2023
38 points (100.0% liked)
Lemmy
496 readers
1 users here now
Everything about Lemmy; bugs, gripes, praises, and advocacy.
For discussion about the lemmy.ml instance, go to !meta@lemmy.ml.
founded 4 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I experienced this on my own little instance as well, although they managed to create 69000+ users before I noticed that something weird might be going on and I got a message from a helpful soul indicating the same thing.
I was pointed to a writeup by lemmy.ninja @ https://lemmy.ninja/post/30492 on how they experienced this and proceeded to clean everything up, and adapted this to my own instance. Only thing to mention of note was that I disabled triggers while doing the cleanup. The DELETE statements on person table was excruciatingly slow to the point where I estimated that time to completion approached 48 hours.
Could you explain what that does and how you did it? I can add that to the post as a note.
Sure. Triggers are normally a good idea as they make sure that data is consistent. Like when you delete a user, a trigger will run to also decrease the number of users by one. But since they run for every row, they can certainly impact performance. Foreign key checks are also implementet as triggers so if your missing an index and the db has to crawl through huge tables of data for every delete (I suspect this was the cause of slowdown during DELETE), that too will affect performance.
While I don't do it often, here is the superuser command I use in psql to disable the triggers before doing any other commands:
SET session_replication_role TO replica;
This is something postgresq uses internally when applying replication data, as it assumes all the data is correct and valid and don't fire any of the triggers or rules that would normally apply when modifying data. As you can see from the name, this is a session setting. If you quit the db session, everything goes back to normal so no data is changed and you don't run the risk of forgetting to change it back when your done.
If you do want to go back to normal operation during the same session for some reason, this gets you back to the default:
SET session_replication_role TO origin;
And thank you for the helpful post about the bot problem in the first place.