I am hesitant to say that I have good news or that I have fixed anything… I honestly don’t know if there is anyone left still using this site after all the problems we’ve had.
That said, I noticed that there were some very recent (as in yesterday/today) commits to the development branch of kbin that look like they address the issues that fedia has been experiencing. I moved fedia to the development branch and reloaded some things, and so far, no error 500’s.
But, caveat emptor, we are on the development branch of early-in-its-lifecycle open source software.
Is fedia.io supposed to be fixed now? It surely is not working well for me. I came back to check after a hiatus of a few weeks.
But, caveat emptor, we are on the development branch of early-in-its-lifecycle open source software.
Living dangerously, eh? You gotta do what you gotta do.
All I can comment ATM is that I’m no longer having the HTTP 500 erros I reported some days ago. Instead the issues I have now seem to do with federation: kbin magazines I’m suscribed to in other servers seem to not be updating (I get a message in Spanish that in English probably translates to something like “warning: this federated magazine is not complete. Explore the full content at [remote magazine link].”). Meaning I can’t ATM interact with magazines not in fedia.io since eg.: the posts or comments I want to reply to are not here.
Which leads me to a question I’ve always had about kbin model of federation but didn’t know how to put in words until now: my current understanding is that “federation” is actually the local instance working on an archived copy of the remote instance content and then sending the new content to be synchronized, yes? If so, is there an alternative model or configuration for federation that actually fetches the remote content when it’s going to be used rather than local users working over potentially incomplete copies?
I’m getting very “git rebase memes” from the current model, so I thought to ask.
My understanding is that when an instance subscribes to a magazine or community, the instance is effectively subscribing to updates and so the local cached copy is kept up to date. That, of course, breaks down when there are issues with federation. I do wonder: are you seeing this with all non-local magazines?
It seems that there is still some problem with certain lemmy instances. For example, lemmy.ml actively blocks kbin based instances (like fedia) and so I don’t think the magazines (lemmy calls the communities) coming from that instance are being updated.
I’m not seeing this with all non-local mags, but I’m def seeing it with all non-local mags that I’m subscribed to. I first thought it was a problem with subscription and tried desubbing and subbing again, but it doesn’t seem to have an effect.
It’s not a very breaking problem when it comes to threads, but it is when it comes to comments, because sometimes even if a thread (post) is mirrored, it’s incomplete and thus I cant participate in a specific comment tree.
I have discovered that the vast majority of my 500 errors have gone away after removing the PWA version off my windows machine. It was using MS edge. Using.firefox, even with highest security settings and ublock origin.
I noticed.first.that in the pages header that magazine was not showing as an option. This sent me.down a.rabbit hole of messing with Edges settings and finally giving up on it and resorting to removing the PWA version.
Once I did that, the 500 errors stopped. Now I only get a 404 trying.to open my profile.
I found/fixed the issue with your profile getting a 404.
If you find any repeatable 500’s, please let me know.
Appreciate it. Although more 500’s popping up. Posted on the.June 29 update thread.
It’s more instances of a problem I thought was fixed: magazines, posts, comments, and users can have associated images. For some reason, kbin thinks it received and created a record in the “images” table, complete with a unique record ID, and associates that image ID with the magazines, posts, comments, and users as appropriate. For some reason, though, that image record doesn’t actually get created. When kbin goes to render a magazine, post, comment, or user, it sees that there should be an image associated and tries to look up the image to get the file name - that query fails and kbin doesn’t elegantly handle that, so an error 500 results any time that happens. I have to fix it by running a series of queries that null out missing image IDs. The bad part is that I am no longer seeing errors that are causing that behavior, so it’s more tricky to debug the remaining issues now, but I am working on it.
