Hi team wave I was redirected from rancher kubernetes channel I have a specific question regarding how to access etcd data in a k3s cluster any help would be appreciated I set up k3s with <https docs k3s io datastore ha embedded|High Availability Embedded etcd> and now trying to access etcd data following <https gist github com superseb 0c06164eef5a097c66e810fe91a9d408|this> instruction However I get inconsistent results between etcd returned results and kubectl What I m trying to do is to get pods count in etcd using `etcdctl get prefix keys only registry pods` and it returns me 7 pods When I get pods via kubectl across all namespaces it gives me 8 pods And the pods names don t match

Hi team wave I was redirected from rancher kubernetes channe rancher-users #k3s

Title

flat-bird-89111

05/25/2023, 9:51 AM

Hi team 👋 I was redirected from #rancher (kubernetes) channel. I have a specific question regarding how to access etcd data in a k3s cluster, any help would be appreciated! I set up k3s with High Availability Embedded etcd, and now trying to access etcd data following this instruction. However I get inconsistent results between etcd returned results and kubectl. What I’m trying to do is to get pods count in etcd using

etcdctl get --prefix --keys-only /registry/pods

, and it returns me 7 pods. When I get pods via kubectl across all namespaces, it gives me 8 pods. And the pods’ names don’t match.

I have 3 k3s server nodes:

# kubectl get node | grep erver
k3d-k3s-default-server-0   Ready    control-plane,etcd,master
k3d-k3s-default-server-1   Ready    control-plane,etcd,master
k3d-k3s-default-server-2   Ready    control-plane,etcd,master

And there’s one etcd runs directly at host port 2379:

# netstat -nplt | grep 2379
tcp        0      0 <http://145.xxx.xxx.xxx:2379|145.xxx.xxx.xxx:2379>       0.0.0.0:*               LISTEN      2229051/k3s server
tcp        0      0 127.0.0.1:2379          0.0.0.0:*               LISTEN      2229051/k3s server

The pods data seems out-dated to me, and I wonder what should

ETCDCTL_ENDPOINTS

be?

Turned out the etcd was connected to another k3d cluster, issue solved!

#k3s Join Slack