The old news.gmane.io server was getting a bit long in the tooth, so I’ve spent the last few days transferring the spool to a new, spiffier server. The server has been down most of the day while I did the final setup, but everything should now be back up again — both the Gmane parts (the NNTP mailing list archive) and the Gwene parts (the RSS to NNTP gateway).
What made this more urgent than usual is, of course, the current wave of sploits — I’ve sorta kinda kept the old server updated in user land, but I never took the actual time to do kernel updates and stuff, so:
That’s not good. So I was just going to update it, but then I discovered that the disk was 92% full, so it was time to move to a new server anyway.
(I was really not looking forward to upgrading the OS and rebooting — it’s been so long that the likelihood of something going catastrophically wrong is pretty high.)
I rented the new server from Hetzner, too, and you can really see how the hardware shortages from a few years ago and then the current AI crunch has affected the offerings: You used to be able to put together just about any hardware combination you wanted. That’s not the case any more.
To get the required amount of SSD (the news server uses a one-file-per-article spool, so you really need SSD) for the server, I also had to get something totally overspecced in the RAM and CPU dept — so it’s a 20 core Intel Ultra 7 265 CPU with 192GB (!) worth of RAM.
Which is nice for the disk cache; sure, but it’s totally overkill.
But whatcha gonna do?
One other reason I was wary of rebooting the old server that much is because I never bothered making all the startup processes completely automated — I just didn’t have the stamina. But now, with the help of Claude, I’ve made startup 100% automatic, so I should be able to upgrade and reboot with impunity whenever there’s a new RCE published (which I’m guessing is going to be a weekly occurrence from now on).
Let me know if you see any wonkiness.
(Oh, and I guess there may be some DNS propagation issues, but I’d imagine all the DNS caches will have flushed by now.)