We are once again about to hit our database quota on the hosting.
This will result in a plan increase, which comes in at an additional $50USD/month.
Currently we're slowing going backwards (income is less than costs), and this increase is going to put further pressure on that.
(how much pressure you ask? doing some quick maths - I reckon we have 2-3 months in the kitty before we're in dire straits)
Times are tough, but if you can afford to donate, I'd appreciate it :)
=> More informations about this toot | More toots from ambassador@photog.social
@ambassador Is there anything practical we can do, like deleting old posts or limiting file sizes? I mean as well as making or increasing our donations.
=> More informations about this toot | More toots from stevenlawson@photog.social
@stevenlawson @ambassador kind of drastic measure, but would doing something like instance blocking bigger instances like .social help at all, since we wouldn't be loading their posts as well as ours?
=> More informations about this toot | More toots from hatebunny@photog.social
@hatebunny @stevenlawson possibly. There's another drastic measure that can be taken as well, which is to nuke data in the database older than a certain date, the problem with that option is, it isn't re-fetched if someone tries to access that information (which seems less than ideal)
=> More informations about this toot | More toots from ambassador@photog.social
@ambassador @hatebunny Personally I wouldn't have an issue with older data being deleted if it meant a secure future for the instance. Out of interest, how far back does the data go, what does it consist of and what cut-off point would you consider acceptable?
=> More informations about this toot | More toots from stevenlawson@photog.social
@stevenlawson @hatebunny The database goes back to the start of the instance (so around 2016ish iirc).
The big scary warning on the setting says:
"Content cache retention period
All posts and boosts from other servers will be deleted after the specified number of days. Some posts may not be recoverable. All related bookmarks, favourites and boosts will also be lost and impossible to undo."
Which is less scary than it used to be, I admit. I might put a poll up.
=> More informations about this toot | More toots from ambassador@photog.social
@ambassador @hatebunny I wonder how often folk go back more than a few months looking for old posts etc? I rarely do - I don't even look at my own posts more than a few weeks back, which probably suggests I should institute an automatic delete option regardless of what happens at instance level. Would pinned posts remain outside any self-imposed or server-based culls? A poll would help gauge reaction, for sure.
=> More informations about this toot | More toots from stevenlawson@photog.social
@stevenlawson @ambassador i’m on board with it. My own posts don’t need engagement after a long period, and i have never gone back to look at faves or boosts - those are for the OPs to feel appreciated.
=> More informations about this toot | More toots from hatebunny@photog.social
@hatebunny @stevenlawson Been doing a bit more digging on this option - apparently it isn't meant for general instances, but for specific announcement type instances.
https://github.com/mastodon/mastodon/pull/27733#issuecomment-1796021422
https://mastodon.gamedev.place/@aras/111351365012735094
This user was somewhat less than impressed:
https://lapcatsoftware.com/articles/2023/10/1.html
I think with a long enough retention period (1-2 years) it might be ok. There'll still be weirdness, and I suspect some users will still be a bit peeved when it happens. hmmm...
=> More informations about this toot | More toots from ambassador@photog.social
@ambassador @hatebunny In reality how many posts or boosts or whatever from 'special announcement instances' does your average photog.social user see or even know about? I doubt it's many. The majority of us probably wouldn't even know this retention feature had been turned off. Or care. On the flipside, if that's all the data it would block, is that enough to make a significant difference?
=> More informations about this toot | More toots from stevenlawson@photog.social
@stevenlawson @hatebunny Oh it's not announcements that get culled, it's ALL external content after the cutoff period. So conversations you've taken part in elsewhere, DMs, any favourites, boosts, bookmarks of toots on other instances.
I don't think many people would be bothered by a 1yr retention, but any shorter than that and I think people might notice and not be impressed.
=> More informations about this toot | More toots from ambassador@photog.social
@ambassador @hatebunny No, I understood that it was 'announcement-type instances'. I still think all that content wouldn't be missed after a year. Just my opinion FWIW.
=> More informations about this toot | More toots from stevenlawson@photog.social
@stevenlawson @ambassador @hatebunny
I personally love those moments when a discussion or post from 6 years back turns up again. Is that worth paying ten times the cost to maintain?
Likely not.
I think the issue of sustainability for instances in the fediverse is so important. If databases are scaling beyond the budgets of small communities then there needs to be some priority given to rethinking some of the core functionality in Mastodon code .
Can this be addressed at a design level?
=> More informations about this toot | More toots from ewen@photog.social
@ewen @stevenlawson @hatebunny It can and should be, but the core dev team has no interest in doing so.
And I guess if you have 1million users on your instance, and 20% through you a $1 / £1 / €1 per month, you probably don't care too much about the hosting costs.
=> More informations about this toot | More toots from ambassador@photog.social
@ambassador @stevenlawson @hatebunny
Starting to see why forks are so popular.
=> More informations about this toot | More toots from ewen@photog.social
text/geminiThis content has been proxied by September (3851b).