urbit
ames: flows continually build up until finally resolving
I don't have much technical detail here other than I've seen constant reports of people having lots of messages come through after poking a channel after some arbitrary time.
I'm seeing this on a fresh moon trying to communicate with another moon of mine on the same machine. These were communicating just fine to start. I left my house, now on a different network, shut one down, brought it back up, and now there's a build-up of messages in multiple flows. Each moon can communicate with my planet just fine.
ship A (ropnec):
ship B (dasseb):
@arthyn Can you share the rest of the output on ropnec from bone 16 so we can see what those flows are? I don't see any particularly weird thing if, for example, ropnec is running a thread on a timer that pokes dasseb. Each of those calls creates a new %ames flow—same as doing -hi ~zod. The interesting thing is that some of the later one we see are retrying so they are not getting acked, I guess because they are not able to communicate at all?
I am seeing something like this with my planet. There are several ships with which my planet have lost connectivity.
Sometimes a bunch of previously sent |his is acknowledged all at once, with connectivity promptly lost again.
Yesterday I have booted up a fresh ship in a different location, only to see to see it lose connectivity in the same fashion.
@arthyn sorry, I was not very clear, I meant from bone 16 onwards (specially for the ones that keep retrying: 48, 52...) so we can see the initiating flows for all the extra ones we see on dasseb—just wanted to make sure what those are.
@mikolajpp Difficult to tell where the problem is but to me this sounds like the galaxy is not forwarding the packets—probably because it's too overwhelmed. When it can finally processed them, all have been queued, and they reach their destination, but then it needs to deal with other ships' packets, so yours are just dropped.
@mikolajpp what planets are those? I'm wondering if they might all be sponsored by the same star and/or galaxy - or a few that are impacted by no-nat and pub ames ping strategies. There's some inconsistency, in regards that, in the core network right now, IIUC.
