The Back End

The server host is doing router maintenance while most of us Brits are kipping, visitors from overseas be warned.

This is for a security patch.

This shouldn't take long, hopefully this will clear a few of the cobwebs from last week's upgrades.

We're going to try to make the leap forward today. There will be some downtime while we attempt to do this. Rollback will be around 3pm if things are not working as expected.

Edit: Finally completed! Sorry it took so long, there was a memory leak I was confusing for the upgrade.

As mentioned here, we need to upgrade Pict-rs to 0.5 for Lemmy 0.19.4 (well we don't strictly need to for 0.19.4, but this is something we have to do eventually). I don't have a reference for how long this will take, but it'll probably be a few hours.

Some downtime on Lemmy will happen as there's some changes to our deployment I want to make, but I'm going to try to keep the instance up while Pict-rs is doing its thing. If it eats too much RAM/CPU though, I may take Lemmy down. Join the Matrix room to stay updated.

Hello, I'm hoping to take the instance to 19.4 in the near future.
Before doing this, I need to take pictrs up a few versions.
This will take a while, so while 19.4 is in the works, it won't happen for a week or so.

Major changes in 19.4 (It's a big-un, and should bring a fair few useful improvements): https://join-lemmy.org/news/2024-06-07_-_Lemmy_Release_v0.19.4_-_Image_Proxying_and_Federation_improvements

If there are any things in this list you're curious about whether we'll be implementing, ask away.

Cheers, GA.

The server seems to be heavily into swap again, so I'm going to do a reboot (and a few bonus patches, because why not).

I might also take the opportunity to start the pictrs upgrade.

If anyone requires replacement entertainment, let me help you find something on TV for an hour or so.

Some of you might or might not know, but I had to do the final stage of migrating the images that make up the back end of feddit.uk

As it was one of those potentially horrible jobs (millions of files on a live site), I didn't want to do it until I had a sec to focus on it fully.
Which meant I was just not getting around to it.

This evening, I got started on duplicating the folder, and doing the initial dry-run of the sync.
After realising the sync would only take 10 minutes or so, I decided that I might as well just stop the instance for a few half an hour, and get it over with. Otherwise it might have been another month until I got to it.
Plus, it's good to restart things occasionally (and makes sure that we can!)

Rough overview was:

• Backup run
• Site taken down to avoid generating new image files
• Current image folder copied to a new location
• Missing files added in at the new location
• Old location moved to temporary folder for safety
• New location moved to old location
• Permissions re-applied
• Site brought back up

And to-do, once it's been running for a few days:

• Delete temporary folder that's chewing up a third of a TB.

That's all folks
Question away if you like.
Light ribbing about it being a retroactive downtime notification is no problem, so long as it's in good humour.

cross-posted from: https://feddit.uk/post/7823042

I won't pin this one, as it should be relatively minor.
A couple of tweaks under the bonnet to apply, and possibly advancing to 0.19.3 if there is time.