Growing pains are to be expected. You're probably aware that some people (myself included) are shifting here [from|in addition to] kbin.social; that extra load probably doesn't help.
Ernest made a post today, yes, but kbin.social has reached a point which demands a next level of administration (from both technical and non-technical perspectives). While I want that project to thrive, there is writing on the wall which unfortunately cannot be ignored.
The server is busily processing the 1,200,000 messages that queued up over the past 20 hours. It’s died 3 times in the past few minutes, so I’m not optimistic about how long this will take
Okay not an error 429 or 500 this time and just some noncommittal Error with no real message in a logged-out UI (that also returns to normal upon refresh) that I once again encountered for the first time.
I upvoted this very post, e-five's comment, and then your reply in rapid succession. The first and last actions went through, but there was a circle indicator beside the "More" option on e-five's comment and the page ultimately loaded https://fedia.io/ecf/3871174?choice=1 and thus the error page. We're both on the same instance and I shouldn't be blocking them or running any script that should mess with upvoting btw
After going back from the error page to submit this comment, hit submit button only to be redirected to https://fedia.io/m/fedia/t/552719/-/comment without the comment having posted. I find the page usually needs to refresh to send a message but this was the first time I ended up on a different page when trying to write something?
Everything about your post is slightly indicative to me of javascript failing, for some reason.
So many actions taken, upvoting / boosting / replying, do javascript actions. However, all of these things have fallbacks in the case either: javascript hasn't fully finished loading on the page yet or javascript requests failed.
As I said this is true for all other actions, so when you vote or boost there are also http GET routes that handle those requests.
It's hard for me to guess more than that, it could be the JS is loading a bit slow and you click before the page finishes loading, or it could be a different error like maybe the request fails. There might also be something else involved, I'd have to double check but mercure might involve itself in dynamic page updating as well... Do you happen to notice if pages are still loading when you take actions, or in fact just browse with javascript disabled? There's also that, last I checked, all the JS files had defer on them so don't block page loads, which is what makes this more likely to happen. Whether that's good or not, I can't say, it's been like that since I first looked but might require more research to know whether it can be split between JS that should block page load and JS that can be deferred
The explanation of what happened makes sense, but indeed it's hard to say how it's happening. My internet speed is decent so I'm certain everything on the page must have loaded by the time I started clicking. I don't disable JS, and the only extension I have active is one for grouping tabs because I happen to access Fedia on my "side" browser, so not even an adblocker. Blocking nothing at the router level either so not sure how requests could have failed :/
Got one just now while hitting next page in my feed.
/?p=4
500 Internal Server Error
Sorry, something went wrong on our end. If you continue to see this error, try contacting the instance owner. If this instance is not working at all, check out other Mbin instances in the meanwhile until the problem is resolved.
@jerry there was a super fast follow up to 1.4.0 to fix a 500 that got introduced https://github.com/MbinOrg/mbin/releases/tag/v1.4.1 (the big warning at the top is to make sure people read the 1.4.0 upgrade docs since this release came in so quickly after it, in case they're running <1.4.0, but you should be all set since you already upgraded)
Got an error 500 the other day when trying to visit /m/theonion@midwest.social, what was strange was that it was not rendered "within" the context of site UI but I was shown a totally white page bereft of any decorations except the error message on the top left (first time I ever saw this).
Got an error 500 just now from replying to a message, it was a typical error 500 taking up the main body column on the site (appeared to be logged out and all), and then returned to normal after refreshing.
EDIT: Just tried out spoiler syntax for the first time to update a new thread and submitting the edit also gave me a regular error 500. Edit didn't go through and I had to go back to resubmit.
Nice, I was just back from being mystified at this comment chain as people seemed to be mixing up the mbin instances, hope the maintenance makes Fedia easier to use regardless of whether folks were planning on landing here lol
this is still somewhat speculative but if you're running this behind a proxy, you might want to configure either nginx realip module or equivalent, or this: https://symfony.com/doc/current/deployment/proxies.html so it could see and work with real user ip, if you haven't done so already (only do one of them, not both)
currently suspecting that this is the big part on why it hit 429 rate limit when it shouldn't be: the rate limiter probably only sees the proxy address, not the users, and so it ended up rate limiting the entire proxy instead of individual clients/users
It has? They had problems with a spammer months ago, but that has been dealt with. If anything, the mods on sh.itjust.works were quicker to remove CP when it came on from other instances, than those instances themselves.
I will lift the ban again. This has happened twice in the past, and each time within ~12 hours, I've had more child porn landing on the server from them. If it happens again though, I will likely leave it that way.
Damn has it been that long already? I saw the moment when it broke, it appeared to coincide with a network outage after a couple weeks of being reliable. It's got to be some kind of simple little bug so I even started looking at the code but didn't get anywhere yet.
Just to follow up on this a bit I took a look at the code to try to understand why this has always been a problem in /kbin/Mbin. It seems like from cursory reading it will try to send out unsub/unfollow requests on blocking magazines/users even if you aren't subbed or following them. Assuming this is accurate, this has some benefits to make sure the remote removes you, but has the downside of calls to the remote needing to be working in order for a block to happen. I'll try to take a look at making it only send those unsub/unfollow requests if you currently are according to local and see if that makes sense to others
Fedia runs Mbin at the moment, not /kbin, issues can be reported on github
but assuming you may not have a github account or posting via the fediverse is easier, you can also try posting mbin specific, non-fedia instance questions, comments, or concerns to https://fedia.io/m/Mdev@kbin.run
fedia
Hot
This magazine is from a federated server and may be incomplete. Browse more on the original instance.