I'm running metallb (image:
quay.io/metallb/controller:v0.11.0) on kubernetes v1.21.5+k3s2 in L2 mode. I have one traefik endpoint running, and it is working, but I frequently get connection refused and no route to host errors. The error condition lasts 5 to 10 seconds, then it works as expected again for many minutes.
arping the endpoint IP during this issue shows timeouts; but even after the ARP entry is back, there are connection refused errors for a couple more seconds.
I think the issue started after upgrading the nodes to Debian 11.
How would I go about debugging this?
Thanks,
Stefan
--
Stefan Bethke <
s...@lassitu.de> Fon
+49 151 14070811