Kbin.social down again? Are people moving to mbin instances (kbin.run, fedia.io)?
Kbin.social down again? Are people moving to mbin instances (kbin.run, fedia.io)?
Link to a list of mbin instances: https://fedidb.org/software/mbin
You're viewing a single thread.
According to that website, Kbin.Social has >10x more registered users (had? looks like it only counts accounts) than all Mbin servers combined.
At this point people need to stop being surprised - whether he is sick or whatever the cause, this is by no means a rare occurrence for that instance.
I think it is good to point it out though. kbin.social is missing from the fediverse observer, but if you have a look at this: https://mbin.fediverse.observer/list you'll see that almost all mbin servers have a >98% recent uptime and a >95% uptime over the whole lifetime of the server. Sadly, fedidb does not have an uptime metric
(yes mine is not up there, because it was offline for a week in september last year)
https://kbin.fediverse.observer/kbin.social there you can go to graphs and see the uptime. The overall uptime is actually still quite good with 95.94%, but in recent months it has been a bit rough
A lot of the time it was technically 'up', but just non-functional/unusable.
Most common for me was just not being able to do anything but look at the front page, couldn't click on anything without errors.
That is an excellent point - I so rarely went there but I thought I recalled that being my experience as well, and yet I wasn't certain enough to say so. It really does mess with the stats if we are trying to use "server uptime" to compare between instances or Kbin vs. Mbin.
well server uptime is usually, and in the case of fediverse observer, coupled to a successful response. If the server spits out a 500 internal server error, that does not count as being up
Kbin still has 3.8k monthly active users: https://kbin.fediverse.observer/dailystats
Curious to see how it will be tomorrow
But that page is for "Kbin", not specifically "Kbin.social" which I note does not appear among the list of all Kbins already - https://kbin.fediverse.observer/list. So you don't have to wait for tomorrow - it's already too late to see its former traffic today.
Interesting: the Active Users Monthly (https://mbin.fediverse.observer/stats) for Mbin is 568, whereas that stat for Kbin was 2280. So even without including the extremely large Kbin.social (well... large in terms of total users, but obviously not active ones bc the service is down, which by definition precludes people being active on it:-), the suite of Kbin instances still seems to have ~4x more active users than the Mbin ones.
I would not have expected that, given the chatter about Mbin being exciting, and I wonder why - potentially historical precedence, if an older server simply has more traffic bc it was created first?
But obviously something more is going on with that data - i.e. & e.g. supermeter.social is reported to have the highest user count among the Kbins, but with only 736 total users, and if you add up all users from all 8 of those servers you get only about half of the 2280 "Active Users Monthly" figure - so I suspect that the activity for Kbin.social is being included in that after all? Otherwise something is very wrong with the extrapolation of "active users", to be more than twice the total ones (one possibility... past active ones vs. a smaller current total of people who deleted their accounts rather than merely abandoned them by walking away without going to the trouble of deletion).
Which would make sense - the website is reporting numbers accumulated over time, and even though Kbin.social is down now, it was not always thus, and it seems it cannot discriminate the history in terms of active users (Kbin.social vs. some other Kbin server I mean).
But that does complicate - possibly even invalidates - trying to compare the non-Kbin.social Kbins vs. the Mbins, in terms of active users.
So leaving active users aside then, I note that the largest Mbin has a ~6-fold higher total user count than the largest Mbin server. Also there are 8 total Kbin instances (aforementioned not including Kbin.social bc it does not appear on that list today), vs. 23 total Mbin instances. It's shaky, but it really does look like the Mbin instances seem healthier than the Kbin ones? (Again minus Kbin.social, which despite monthly active users seems by no means "healthy" to me?)
This ignores things like possible hyper-focusing on specific niche topics so a deeper look would involve how many communities are there, and perhaps traffic patterns like do people actually comment in those or is the server mostly just a base from which to access the Fediverse at large (which may not be a bad thing at all? just a bit different), etc.
so I suspect that the activity for Kbin.social is being included in that after all?
I think so, because https://fedidb.org/software/kbin shows the MAU for kbin.social, and it's 3,843
I think the instance is not showing because it is not online at the moment... here you can see the stats of kbin.social on fediverse observer https://kbin.fediverse.observer/kbin.social
Regarding the amount of instances and the rather small users/server: a lot of the mbin instances used to be kbin ones (like you can tell by their name) including mine. So we did not start with one central instance that all the users went to, but with a lot that already had a small number of users. And the project itself is not that old, not even a year (we start in September or October 2023). I'd say we really only have one hyperfocused instance and that being rimworld.gallery
And to be clear, hyper-focused isn't "bad", just not the same as someone wanting to join a more general-focused one.:-)
Thank you for sharing some of that back story.
Really surprised by that. Granted, I logged in yesterday to check and that makes me part of that statistic but it's like 90% spam there and many places do not federate properly. The kbin.meta magazine, who clearly has newer posts, do not show anything but 7 month old threads from within kbin.social itself. Super weird, but ultimately unusable.
I'm not sure of the accuracy of the user count for kbin, because account deletion requests, at least for kbin.social, are not being processed. <edited>
I believe it, but I already did not trust those numbers for a different reason - e.g. I abandoned my account there six months ago to come to where I am at now, so technically I have an account and yet I've been there like 6 times since then and commented or interacted fewer than that.
Still, the total user count represents a "high mark" that it had once reached, and the Mbins collectively still seem far away from that. But good point, b/c how many accounts are e.g. alts or deleted from Mbin successfully but from Kbin that request gets ignored.
"Activity" would be a better measurement. Down below in some of the other replies we looked into that, and I think technically Kbin.Social is still fairly active, more so than the Mbins, but overall the Mbins are obviously in a healthier state with fewer of these insanely long (weeks-long) outages.
Btw, in my link above (for "sick"), Ernst mentioned that:
The care of the instance will also be handed over.
So it looks like things will change at Kbin.Social regardless of his health & life issues.
I hope things turn around.
I for one hope that he goes back to what he seemed to enjoy the most: writing the code. Let someone else handle the admin duties, which he mostly abandoned anyway. We would get the non-Lemmy codebase enhanced, while he would get the fun of chasing his passion:-).
Is that likely to be a large percentage?
The point is there are those who, like myself and others, who requested account deletions on kbin.social. And they have not been processed. Over a period of time those numbers add up. Then you have whole instances that moved (kbin.earth etc)
That doesn't wound like a very relevant point at all.
A relevant point would have been something that had a significant impact on the thing we were talking about, which is why I asked.