Seems .world and .ee federation are broken

submitted by πŸ‡¦πŸ‡Ίπ•„π•¦π•Ÿπ•₯π•–π••π•”π•£π• π•”π••π•šπ•π•– edited

Is it a .world or a .ee issue?

Seems .world to .ee is broken so i wont be seeing anyone's comments from .world and the rest of the fediverse probably wont see this.

Heres ya issue: https://grafana.lem.rocks/d/bdid38k9p0t1cf/federation-health-single-instance-overview?orgId=1&var-instance=lemmy.world&var-remote_instance=lemm.ee

Edit found more info

Log in to comment

28 Comments

Nothing4You

it's not just lemmy.world.

of the larger instances, the following have trouble sending activities to lemm.ee currently:

  • lemmynsfw.com -> lemm.ee: 2.81d behind
  • sh.itjust.works -> lemm.ee: 1.04d behind
  • lemmy.world -> lemm.ee: 22.5h behind

i pinged @sunaurus@lemm.ee on matrix about 30h ago already about the issues with federation from lemmynsfw.com, as it was the first one i noticed, but I haven't heard back yet.

Steve

Seems like theres some widespread DNS problems again today

r00ty

Wait, how do they get that data remotely? I was looking at my instance vs world and I saw there's like the +1 hour from a week or so ago when I upgraded to latest mbin lol.

I guess they're looking at common activities and when they appear on each?

Nothing4You

lemmy has a public api that shows the federation queue state for all linked instances.

it provides the internal numeric id of the last activity that was successfully sent to an instance, as well as the timestamp of the activity that was sent, and also when it was sent. it also includes data like how many times sending was unsuccessful since the last successful send. each instance only knows about its own outbound federation, but you can just collect this information from both sides to get the full picture.

there is also https://phiresky.github.io/lemmy-federation-state/site to look at the details provided by a specific instance.

r00ty

That makes sense. So it's showing me world's federation with me and not the other way (since I'm not sure such info is available on mbin)

Nothing4You

pretty much, yeah. lemmy has a persistent federation queue instead of fire and forget requests when activities get generated. this means activities can be retried if they fail. this allows for (theoretically) lossless federation even if an instance is down for maintenance or other reasons. if mbin has a similar system maybe they could expose that as well, but unless the system is fairly similar in the way it represents this data it will be challenging to integrate it in a view like this without having to create dedicated mbin dashboard.

r00ty

We can see it ourselves. We use rabbitmq for incoming (and maybe outgoing, it's been a while since I looked at how it is) federation. So, you can see the queues there. For incoming (from rabbitmq) and outgoing there are also queues (symfony messenger) and these handle failures and can be configured and can be queried.

After the upgrade I just took the default configuration again (because it seems queue names changed). But I used to have various rules setup in rabbitmq for retries and it took a fair few tries before the messages ended up in the proper "failed" queue (which needs manual action to retry). Some items you eventually need to clear (instances that just shutdown, or instances that lost their domain for example). They will never complete.

But it's not exposed in any way to my knowledge. Well unless people have their rabbitmq web interface open and without login of course.

Cris , edited

Huh.

I can see the post from my .world account

muntedcrocodile

Yeah but u cant see this comment from my .ee account

jordanlund

I know there was a problem where federation was lagging by a couple of days, but I thought it was fixed? I'll hit up the Admins.

muntedcrocodile

Nar its lagging now. Ps Can only see this cos im on a .world alt

Daemon Silverstein

and the rest of the fediverse probably wont see this.

As a The Lemmy Club user, I can properly see the post. Federation seems to be working okay.

TriflingToad

I'm seeing this 12 hours later lol

r00ty , edited

Looking at incoming request. .world is working OK for me. They seem to be batching stuff like I'll get nothing for 30 seconds, then over 3 seconds like 50+ requests.

Of course I don't know if their queue is backed up and I'm getting delayed stuff. I'd need to stop processing and look into the incoming queue to see what they're sending.

Bit of an edit. Looking at incoming again I can see under newest items, an entry from world that was 11 minutes old. Oh I have an idea. I'll see if this edit gets there in a timely manner.

Spoiler alert, it was instant.

Oh ignore me. It's specifically between those two instances I guess.

Aparently its between big instances and lemm.ee only. All of said instances use cloudflair. I suspect that cloudflair has blocked/rate limited the larger instances from reaching lemm.ee.

r00ty

I think it must be hit/miss. Because I think those edits I made would have gone from my instance to world and then from world to .ee, and it was happening within seconds.

So, presumably random stuff is being dropped or delayed?

Yeah ur right. Everything going on here must go through .world as it is the community. Which makes it even weirder that we dont have .world users federaring out.

My understanding of federation is ur events get sent to ur instance (kbin.life) ur instance sends those events to the instance of the relevent community (lemmy.world) the instance of the community sends that event to all instances subscribed to that community (lemm.ee, kbin.life, etc) my instance receives that event then notified me.

So this conversion is goibg through .world just that .world users events seem not to be gettibg sent to .ee

I think we need someone smarter than both of us to fix this

SagXD

I hope it get fixed soon. Currently using this alt.

originalucifer

its about 25% of my instance traffic. its noticeable, but theres definitely a shift towards spreading content out.

SagXD

It is.

jordanlund

Admins have been talking about it, it looks like it has something to do with banning/unbanning users. Sometimes a ban/unban doesn't want to propagate to .ee and that causes a logjam somehow.

Smarter people than me are looking at it!

If its that bloody bot that bans u if ur comment karma is to low thats gonna be funny. I got banned/unbanned like 10 times in an hour cos my comment karma was fluctuating near the bots threshold.

Blaze (he/him)

Til there is such a thing

Ech

That thing is such a nuisance. Makes the modlog nearly unusable.

w3dd1e , edited

I see it? Maybe I just haven’t noticed but I think .world is fine for me?

Edit: nvm you might be right. I just checked my posts in !futurama@lemmy.world. They normally get 60-100 up votes but the last one only got 3.