![](https://aussie.zone/pictrs/image/2644a4fe-cce3-4126-b80b-40e6369976d1.jpeg)
![](https://aussie.zone/pictrs/image/1d0e268b-6158-4726-a025-e09109d5d409.jpeg)
You've got a way with words... are you a writer? :)
I also have backup accounts on these instances:
https://beehaw.org/u/lodion
https://sh.itjust.works/u/lodion
https://lemmy.world/u/lodion
https://lemm.ee/u/lodion
https://reddthat.com/u/lodion
You've got a way with words... are you a writer? :)
Prior to restarting the lemmy service, this showed over 2k instances as "lagging". Shortly after restart, its dropped down to single digits.
I'll leave the hourly restart going for now, it should help with federation issues... and should help with memory leaks too.
I've given it a kick now, and re-enabled the hourly lemmy restart. But it looks like this needs to be fixed by the devs.
Yeah, another reboot... have updated to lemmy 0.19.1 also. For some reason we're hitting memory exhaustion on the host now... never did before. I'll see what I can figure out.
Someone has misunderstood my previous comments on this topic: https://aussie.zone/post/298893
A quick server reboot and it seems to be back. Looked like my postgres config caused some memory issues, which killed off back end object storage... and pictrs barfed.
pictrs seems to have given up, and is refusing to run... new thumbnails etc are not being created. I'm looking at it.
I've not seen that myself, will look into it tonight.
Lots of back end improvements... on the user facing side 2FA enrollment improvements, and highlighting of recent comments. I've not noticed anything else, but I'm sure its there.
That would be my guess. There have been pretty major changes to the API as I understand it.
Noticed the same. Seems to be cosmetic so I'm ignoring it for now 🤷♂️
I didn't see anything bot related from my skim of the change notes... should be fine.
You've got a good memory. Yes, the new process is slightly different so the old guide isn't relevant. Also its not as important anyway, enrolling no longer enables 2FA before you have a device capable of generating TOTP codes.
This was deliberate on the part of the devs. Old 2FA enrollment was finicky and error prone, locking many users out. So they've disabled it for everyone, forcing re-enrollment to enable it.
Edit, nope... postgres docker memory issue... seems to be fixed after adding "shm_size: '1gb'" to the docker postgres compose stanza.
Spotted this one myself, but its not isolated to a user's own profile. Suspect its a back end lemmy issue... will investigate.
Ok, I've run a DB backup, updated docker-compose and pulled new images. I'll do a restart some time today.. and hope for the best.
I'll take a look at it. Short version... last time I checked it was still an RC, which I didn't want to play with.
Updated