Publishing our data backup schedule
As part of accountability to our members, we should make public how frequently and in which manner all social.coop data is backed up.
Nathan Schneider Wed 11 Apr 2018 5:31AM
The wiki (which has been experiencing erratic downtime) might be a
good place to share such technical materials as well, since they may
not be of interest to people seeking basic information on the About
Erik Moeller Wed 11 Apr 2018 5:38AM
I agree the wiki is a fine place for really going into the details. The high level info (bullet point level) about backups, at a minimum, does IMO belong on the "About/More" page, both because it's a reasonable question for any user to have, and because it's a conventional place to look for it (all Mastodon instances have it, while not all instances have a wiki).
Neil - @firstname.lastname@example.org Mon 16 Apr 2018 10:29PM
Agreed, this is important. I agree it doesn't need to be hugely public, but certainly discoverable. A summary on 'About/More' seems a good place, along with a link for more detailed information (if needed) to the wiki.
In addition to backup schedule, some documentation of a disaster recovery plan would be useful. At least documented on the wiki. (It could be that restoring from backup is pretty straightforward with Mastodon - don't know currently.)
Victor Matekole Tue 17 Apr 2018 9:16AM
@erikmoeller5 Let's discuss the technicalities of this on the call. It would be great if you could lead on this initiative, with my support of course.
Erik Moeller · Wed 11 Apr 2018 5:28AM
The "about/more" page is a good place to publish technical information about an instance (see https://mstdn.io/about/more for a positive example of how to use this page for technical details). The most crucial technical detail is whether, how frequently and in what manner data of our instance is backed up. I don't actually know the answer myself.
It's not an uncommon occurrence for major data loss to occur when self-hosting complex software. This can happen in the course of upgrades or other complex system administration, but it can also happen due to hardware failures. For a user to choose what instance to use, one of the most important questions they have is: is my data going to be handled responsibly?
Ideally, for all services (i.e. social.coop itself, the wiki, and other self-hosted services), we should have both local and remote backups produced frequently, perhaps nightly. But whatever the current state is, I suggest we document it publicly.
I'm happy to help with this if someone wants to just give me a verbal download; I can then turn it into a document.