Older news and announcements
Long story short: by trying a bit too hard to clean up the server after the upgrade, I broke the system badly enough that it wouldn't boot again despite my best efforts.
(Just to be clear, the upgrade itself went just fine and you shouldn't be afraid of upgrading your own Debian based on my little mishap.)
In the end, I reinstalled Debian 11 from scratch and re-imported all user files. This had the added benefit that I did end up cleaning up a decade of cruft dating back to much earlier versions of that server.
There should be no noticeable difference to our users, except, again, a new SSH server key, whose fingerprint is now:
SHA256:p7jp2y5Hs+0FrPvujIsS/oEtWB7jcShLFI4wn7vTAiU
All my apologies for those two days of downtime — although averaged over the last decade, we're still around 99% availability. Nevertheless, if you have stuff on hisoka and it looks broken, or if something needs to be reinstalled, do yell at me!
I promise once all is working fine again, I'll try very hard to not touch anything for the next ten years.
— Kefen
The upgrade to Debian 11 went well but the ensuing cleanup did not. All user files are safe and the system is being reconfigured from scratch. All my apologies.
Please stand by!
— Kefen
After a decade of everything working perfectly fine, and after saying I was going to leave it that way, I have decided that I want more performance out of my server.
Kimsufi had interesting offers again, so I got us a KS-LE-1 server with 32GB RAM and 4TB HDD, as well as a beefier CPU.
Everything will soon be copied over to this new server and you shouldn't have to do anything, there will just be a complete interruption of all services on the current server while it happens and until new DNS entries propagate.
You will get a warning the next time you SSH into your account. The new server should advertise a new fingerprint for the host key, which you should put in your known_hosts file instead of the old one:
SHA256:Jag4aUwi/maRPfGRjc3PjN3uvGd0oKWvT6nY+LNK6Lw
As usual, sorry about the inconvenience. This is for the greater good, though.
— Kefen
For the longest time — it's been enough years that I can't tell how many it's been — the old website had a "planned interventions" section with only one item: "website overhaul and better management of our users".
Well, that's taken care of!
Other than that, Stuff just Works, and I hope it'll last another decade.
— Kefen
I got alerts from mdadm this morning indicating, via a pretty terse DegradedArray message, that /dev/sda had died.
I want to praise OVH for how fast they fixed the issue. Within a few hours, the fault was confirmed, the hard drive physically replaced and as of this post, the RAID array is happily rebuilding.
Apart from the fact they had to reboot the server around 15:47, there were no other issues. But as usual, do yell at me if something seems missing or corrupted.
— Kefen
― Zach Weinersmith