THIRD UPDATE:
Hopefully this is the final one! I have managed to salvage the original image upload database (which was not easy, considering that it’s a very niche database format 😅)
All missing avatars and banners should now be restored!
SECOND UPDATE:
It seems that not all user uploaded images survived the migration. I’m really sorry, but if your community or your profile is missing their avatar, I would ask you to just re-upload it.
Why this happened is not clear yet, and while I do have a full back-up of our uploaded image database, I can’t just restore it now, because post-maintenance image processing has already begun, so restoring the backup fully would just overwrite the newer uploads. I will dive in and see if I can salvage any of the old data from the back-up, though.
UPDATE:
The database upgrade has been completed! Overall it took a bit longer than I hoped, due to some issues with importing the database backup, but everything should be properly migrated over now and running smoothly.
Some users may experience icons not loading. This is a cache issue and should fix itself in a little while, but doing a “hard” refresh in your browser should fix it immediately (for example: Ctrl + F5 for chrome on Windows).
I’m excited to see how the new database holds up, so far the performance is looking much better, but I’ll definitely keep monitoring it!
Original message:
I woke up today to a pile of e-mails notifying me of new donations from users of lemm.ee. I am super thankful to everybody who has decided to donate to me, as I am now able to take immediate action for infra upgrades thanks to the increased budget.
Over the past few days, our infrastructure has begun somewhat struggling with load. While our local user count is not super high by any means, our database still needs to handle incoming posts and comments from the rest of the federated network, and this is where the majority of the load is coming from.
I am going to put the community-provided funds to good use without delay to double our database specs, which should hopefully give us some headroom in terms of performance. This upgrade will require downtime, as I will be migrating the existing database to a new machine completely. The maintenance and upgrade will begin when this post is 3h old.
Future database hardware upgrades may be possible without downtime as well, but for now I would rather prioritize a quicker fix over an extended preparation period, and this means doing it with downtime.
I have to say, this is an interesting way of announcing time. Most of the time I’ve seen 1600 GMT or EST.
This form of time doesn’t need any calculation and is universal to all. Quite innovative.
I’ve seen that before on reddit
Godspeed with the update, but don’t sweat it if you run into delays. Everyone around here is very appreciative of your time and effort, thank you.
Thanks for the transparency!
👏 Great work and transparency @sunaurus@lemm.ee ! You all have great instance admin. He definitely deserves the support.
16:00 CET, 10:00 EST, 07:00 PST for us people that prefer absolute times ;)