[Usernetes] kube-apiserver fails to connect to etcd: dial tcp 127.0.0.1:2379: connect: connection refused
#65
Comments
|
curl result is weird for https://10.100.201.100:2379 $ nerdctl exec usernetes-node-1 curl -k https://127.0.0.1:2379
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0
curl: (7) Failed to connect to 127.0.0.1 port 2379 after 1 ms: Couldn't connect to server
FATA[0000] exec failed with exit code 7
$ nerdctl exec usernetes-node-1 curl -k https://10.100.201.100:2379
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0
curl: (55) getpeername() failed with errno 107: Transport endpoint is not connected
FATA[0000] exec failed with exit code 55 |
|
Trying to isolate the issue from Kubernetes, but apparently the issue can't be reproduced with a simple etcd-only container $ nerdctl run -it --rm --annotation nerdctl/bypass4netns=1 ubuntu:24.04
root@f601efbaeb12:/# apt-get update && apt-get install -y etcd-server curl iproute2 jq
root@f601efbaeb12:/# ip=$(ip --json route get 1 | jq -r .[0].prefsrc)
root@f601efbaeb12:/# etcd --advertise-client-urls=http://${ip}:2379 --listen-client-urls=http://127.0.0.1:2379,http://${ip}:2379 &
root@f601efbaeb12:/# : 404 is expected here
root@f601efbaeb12:/# curl http://127.0.0.1:2379
404 page not found
root@f601efbaeb12:/# curl http://${ip}:2379
404 page not foundProbably the problem is about that the etcd is running in an etcd pod container inside the node container |
|
I've never tried bypass4netns with nested containers. |
|
Workaround: disable bind (it is already fast without bypassing) |
|
I confirmed the issue. The below log shows the situation then I created netns and executed iperf3 in a container with bypass4netns. bypass4netns wrongly bypasses the iperf3 socket. |
I'm trying to run Usernetes (single-node w/o VXLAN, as a baby step) with bypass4netnsd, but kubeadm fails:
Looks like kube-apiserver is failing to connect to the local etcd due to
dial tcp 127.0.0.1:2379: connect: connection refused,although the etcd process is running with
--listen-client-urls=https://127.0.0.1:2379,https://10.100.201.100:2379.Version
Logs
The text was updated successfully, but these errors were encountered: