Starting today, I noticed some of my comments and posts aren't being visible to me when I am logged in. According to the RSS feed, logging out and checking it through an incognito tab, the posts are successfully posted on the instance and others', however.
Take this post for example: https://compuverse.uk/post/822
Here you can see I am able to view my post:
But when looking at my profile, said post is not visible:
Same applies to several recent comments I have made. Am I missing something, or is there something wrong with the instance at the moment? When logging off, I can see the post just fine on my own profile:
This is also the case when using the Android client (Jerboa), and when trying to hard-refresh pages etc.
I don't have an answer for you but some data that might help your debugging:
๐ I'm Cameron, the Admin of the instance in question.
It's a very odd situation, I'm entirely new to this whole Fediverse thing myself, but, trying my best!
I noticed in my log, literally just after @TopHat@compuverse.uk posted their comment, this popped up:
I'm unsure as to exactly what this is referring to, as the comment was made from my instance, but it appears the instance believes it came from elsewhere?
It's very odd and I can't really explain it I'm afraid.
I'm beyond the limit of my knowledge and don't think I can be any more help here.
The one thing I do notice, is that the language of that error message seems to suggest that your own instance is regarding some of this traffic as invalid traffic from a federated peer? I'm zooming in on
cannot accept local object from remote peerhere, which makes it sound like your lemmy instance is misidentifying itself somehow. This makes me think about potential DNS mismatches, are you running lemmy and lemmy-ui in docker containers? Could lemmy-ui be identified by one name in configs, but reverse-resolve to a different name from within the lemmy docker container? I'm speculating wildly here and I'm not sure what I'd even do with the answers to these questions. Just thinking out loud.Maybe someone else who has actually admin'ed a lemmy instance will weigh in. There's certainly a lot of useful data in this thread that might allow another lemmy admin who has seen this to chime in. You could also consider cross-posting to https://lemmy.world/c/lemmy_admin@lemmy.ml for more visibility.
Actually, your suggestion might indicate something important. The domain of the instance did change, with the old one giving 404's to get it off the list. Although Cameron did check the configs and didn't find anything wrong with them.
@Cameron@compuverse.uk Worth to cross-post to the admin channel, with the context of having switched over domains?
Technically the domain of this instance didn't change. This instance was setup from scratch, under
compuverse.uk.However, a different instance did run under the same domain for a brief period of time, which was removed, and completely replaced by a brand new fresh instance.
It is possible that nginx or some other http proxy is forwarding requests via some borked virtual host or url rewriting config, such that requests are arriving at Lemmy tagged with the old name?
Perhaps the affected user could clear some browser caches as well to ensure there is nothing funky going on in their specific browser.