Time to come back, it is working again. Lucky break π
Kris
Slrpnk.net is unexpectatly back online π₯³
Nein, die alternativen, bereits quell offenen Treiber waren eh besser und jetzt halt die einzigen.
Ah, right they just swapped out without migrating anything. Thats not really what we are aiming for, but their comment confirmes that remote subscriptions might get lost.
I would need to check at database level, but I guess on the subscribing side it is only a list of communities to display? The real "work" for subscriptions happens on the remote instance to decide with activities to send.
But that indeed raises the interesting point on how to retain subscriptions of remote users. Anyone got an idea how that is handled on database level? Is it just the same list but for the non-local users that are mirrored in the database?
They actually migrated stuff from Lemmy? I thought it was a new instance?
I think we will try to come up with a theme for Piefed with similar colors.
The backups are most likely fine π
Sure, that seems reasonable. We will add that when the instance is back online.
If I forget, please remind me. For sure we will not just ignore you π
Regarding the price and all that, maybe that is an interesting discussion to have with the slrpnk.net community? And it would also be good to let slrpnk.net users know how the instance is being ran, what hardware, where, etc, so users can be an educated decision on whether they want to see an instance or not? Maybe providing that info on the sidebar would be nice?
We have been very transparent about that and documented that in the wiki which is linked in the sidebar. But it seems few people ever read the Wiki and having it integrated like it is with Piefed would probably help.
I agree that more admins with remote access would be good to have, and we have been working towards adding more (there are already more than one, who are not directly involved with slrpnk, but rather the mother project f-hub.org), but physical access will always be limited by the location (Azores) unless we move to an entirely different data-center (which in turn would make it hard for me to access the physical servers).
Yes, I also suspect that at a certain scale it would make a difference, but with a single Postgres database and no advanced clustering or so, the real-life performance metrics of a medium sized Lemmy instance strongly point to Postgres being the bottleneck and not the Python or Rust codebase of the rest of the software.
We managed to resolve the issue and I have many ideas how to avoid this in the future π€