Mastodon isn't sustainable.

Even on what is mostly a single-user instance like mine it uses around 10GB of disk space per month. The database stores 348.389 statuses.

If we want Mastodon to succeed, we have to solve this.

Scare tactics aside, there is some discussion on GitHub about this and the consensus seems to be that removing remote content after a certain time could work.

But this is a pretty hard problem that can affect all mastodon users and @Gargron himself has said that he needs help figuring out what to do.

@sluglife what are your thoughts on content having an expiration date? i personally prefer it over the "store everything forever" approach to social media

@amydentata I think we really have to consider what we want Mastodon to be like in this regard.

Allowing each instance to set it's own expiration date might be interesting and lend more weight to a users choice of instance.

But then again, we can never really guarantee content deletion once content has been federated to other servers.

@sluglife it's true! "the internet is forever" and all that. but at the very least, locally to the one instance doing it, i support automated archive deletion

@amydentata I mean, combined with automatic deletion of remote content (which is really what is needed to cut down on the disk usage), volatile content probably could work well for a lot of instances, particularly the fast-moving and meme-y ones.

@sluglife cloud storage providers solved this for, at least in the short term. other instances solve this by expiring old statuses and media. please don't give sensationalists fodder—this is how that godawful daily dot article gained traction

@sluglife there are also cleanup tasks you can do to remove e.g. autogenerated media previews

but on a fundamental level: yes, if you offer a service whereby people can post media and text and no way for these posts to be removed at the same or greater rate, then you obviously will grow storage requirements without bound????

@io @io Yes, that post was intentionally a bit sensationalist but the fundamental issue that still hasn't been solved isn't that people are posting too much on my instance, it's pretty much just me here and I hardly post any media, the problem is all the remote media that my server is storing with no system in place to automatically remove it.

Feel free to correct me if this does exist but afaik is tracking this with no solution so far.

@sluglife there's actually a tootctl command to remove remote media that's n days old

tootctl media remove

set this to run on a timer and you're good to go!

n is configurable in the main masto config file

@io Thank you, just ran a dry run and this seems to do exactly what I'm looking for!

@io Also could you point me towards how to run these cleanup tasks? I can't find anything in the new docs or tootctl and the rake tasks have all been deprecated, haven't they?

@sluglife yeah definitely, try reading through `tootctl help`. has a systemd timer that runs the following:

ExecStart=/usr/bin/docker-compose run --rm web bash -c 'tootctl media remove; tootctl accounts cull'

we also delete media previews once in a while, but after moving to S3 we have enough storage that we don't need to do that so often

Sign in to participate in the conversation

Smurpspæk Mastodon instance, home of @sluglife and associated cryptids.