Instance Crashes and Migration


#1

When the admin of witches.town decided to shut down the users all had plenty of time to download their data for migration purposes. But what if your server just goes offline without explanation? Obviously it’s a bit late for that download. And yes, it might come back up tomorrow, but long outages do not inspire confidence.

This seems to be a downside to decentralization. Supposedly copies of my posts exist in caches on most of the instances out there, but I can’t just copy all that to a new instance and move on. If my instance doesn’t return, or has lost all my data, I’m just SOL.

Also, does anyone know how to find out the admin contact info for an instance once the site is down? (linuxrocks.online - FYI)


#2

Simple for linuxrocks.online its me :smile: LinuxRocks.online has no plans to go anywhere at this time. We had a outage with our DNS provider and in loo of this we are bringing DNS in house so to speak. I have two personal DNS servers that are in separate datacenters I will be moving it too, We also keep offsite backups of the database and mastodon folder in the other data center as well in case of drive failure. There is no way to stop every outage hell even Amazon Github and Google have had their fair share but what we can do is learn from each to minimize outages.


#3

Someone was spitballing about solutions to this the other day. Though no one but me jumped in, so it didn’t go anywhere. Seems very nessessary though!


#4

I’ve used a google cache of the /about/more page for this before. Web archive should also work.

Yes, indeed. Though to be quite fair, if Twitter went down tomorrow, you’d have the same problem. And if you were using a pure p2p network but lost your private key, that’s game over too. In any case, a couple things: Importing data will likely become a possibility in Mastodon in the future. There’s also an interesting approach to having live backup accounts called “nomadic identity”, implemented by the Hubzilla project.