dpc
@elsirion @m1sterc001guy Thoughts?
FWIW. We are now printing warnings etc. when commit was not called, so I don't remember last time we've seen a bug caused by lack of commit_tx, or even me...
IMO both server and client need to do retries. Server because it should not be flaky, client needs to take care of networking errors anyway.
> Since anything we need to guarantee should be retried on the client anyway because the network might fail, server side failures should not lead to any flakyness. If the...
The logs reported in in `mainline` accompanying the issue which was the reason to create this issue are are know issue https://github.com/pubky/mainline/issues/61 in the `mainline` crate, but the relevant code...
Given the scope of synchronizing `rand` crate to upgrade `mainline` and `pkarr` for iroh, I don't think it will upgraded. For now maybe we can offer ability to disable DHT:...
@Nuhvi At the moment, our main problem is that stable-ish Iroh release (1.35) is using quite old Pkarr version, and we won't be able to switching until well after the...
I don't think there's much to do here anymore. By default we disable dht initialization in the old iroh stack, and the issue is gone.
Kicked off a backport of https://github.com/fedimint/fedimint/pull/7935
I haven't seen it on master since merged, so I think this will fix 0.9. Will reopen if needed.