I kept my #GoToSocial server going for about four months. I just killed it. It was in the Oracle Cloud, and the upgrade of #AlmaLinux from 9.2 to 9.3 was taking a long time. I thought it was hanging, and I hit ctrl-c. That stopped a "scriptlet." I reran the upgrade, which said it completed. I should have know then to remove the new kernel. I rebooted, and that was it.
I hadn't yet set up backups for this server because I considered it experimental.
I can lean into this Mastodon instance (ruby.social), which is very reliable, while I take a break from running GoToSocial.
I never got all that comfortable with the Oracle Cloud. Its biggest attraction is that it's free. I hadn't gotten around to setting up all those sysadmin things that needed to happen.
Now that my Raspberry Pi 4B is set up and running the Caddy web server, I can eventually start a GoToSocial server with SQLite (instead of Postgres, which I used in my last 2 builds).
@passthejoe This is the worst thing about free. Like, you could set up snapshots, but I think you only get a couple. You could then export to object storage, but then… etc
If you terminated without destroying the boot volume, I can help you get at the data from another instance. I’ve had to do terrible things to inspect broken boot volumes ;-)
I was able to pause GRUB with Oracle Cloud's serial console and boot into an old kernel.
Barbaric as hell. That's my review. I removed the latest kernel and then upgraded it. I'm not sure it will work, so I'm not rebooting for now. I'll wait for a new kernel, upgrade that and THEN reboot.
So for now my #GoToSocial server is back among the living.
Time to figure out a backup plan. I've already started. In my mind, that is.
The "trick": Once you've started the Cloud Shell, reboot your instance (click the box to reboot immediately) and make sure you click in the shell window so your browser's focus is there and your keyboard commands will be detected by your instance.
This all started when I was doing a dnf upgrade, and it was taking a LONG time. I thought it was hanging, so I hit ctrl-c. Unfortunately it was in the middle of upgrading the kernel, and the new kernel was rendered unbootable.
I'm sure there are better ways of upgrading a Linux server that don't depend on you sitting in front of a terminal on the other end, and I plan to use them.
Add comment