Wish me luck fediverse, tonight I'm migrating whitespashe.uk from docker.

Follow

Migration failed. Whitespashe.uk is running on a damaged database. If there are any that could give me a hand trying to fix this, I'd be really grateful.

@GigaByte4711 Do you have the error log from the failed migration? What versions were you coming from / going to?

@debugninja
I'm trying to move off docker, and to a normal instance.

The database won't dump, due to an error in the statuses table, so I can't migrate the table.

Here's my discourse post on it: discourse.joinmastodon.org/t/u

@GigaByte4711 Have you managed to fix your issue? It seems like it's postgres related, the only thing I could find was this post: stackoverflow.com/questions/51

I've never experienced it myself, so I don't have any specific tips, but check out the postgres page on data corruption before taking any action: wiki.postgresql.org/wiki/Corru

@renatolond
Afraid not. I read that same issue, and the corruption guide. I've got a FS-level backup, but ad I'm not really proficient in postgresql, I've not been able to make much headway. I'm off work this week, so I'll probably end up doing my troubleshooting on IRC then.

Whitespashe works fine on the damaged db, but I just can't make DB backups. I've been relying on VM snapshots and nightly lun snapshots for my backup options.

@GigaByte4711 If you have a chance to update the minor on your postgres image it might be a good idea, assuming it's a postgres bug it could already solve the issue for you (if you are in 9.6.x to upgrade to 9.6.12, or if you are in 10.x to update to 10.7)

When I had an index corruption regenerating the index solved it, though I had to manually erase some dup entries that were created because the index was a unique.

U could try regenerating the indexes to check if it changes something

@renatolond
I'll try the reindex in a second, see if that helps. And I'll take a look at the minor version options.

I'm pretty sure the corruption came as part of a power outage and disk failure I had early on, but as the database came straight back up and started working, I must have overlooked it.

@renatolond
Image update worked, now running the latest alpine docker image for 9.6.

Unfortunately the reindex errors at the same point with the original error. πŸ˜₯

@renatolond
I tried to follow along with that, but attempting to delete the row returns 'DELETE 0'

Unfortunately, it doesn't say why it didn't delete the row. :(

@renatolond
Posting again because of snapshot rollback:

Yeah, vacuum fires the same error as the other attempts. This week is gonna be filled with frantic IRC-ing trying to get this sorted. The only other thing I can think of is truncating the statuses table, but as it's linked to other tables, they get truncated too.

I effectively nuke the posting history of my instance, and I'm not sure that's what I want.

@GigaByte4711 Can you paste on a gist or something like that the commands and responses you got from trying to delete that bad id?

@renatolond
Sure, gimme a bit, gonna switch to my laptop, rather than ssh from my phone.

Sign in to participate in the conversation
Whitespashe Mastodon

This is a mastodon instance that allows users to share ideas and participate in discussions. Our code of conduct and extended information can be found: Here!