| 17 Apr 2024 |
 DoofusCanadensis | strikes me as immature and unprofessional, but w/e | 21:21:52 |
 sudoer | What do we like for container runtime these days? I'm trying to kubeadm init a new cluster, it's using alma 9 with containerd, kube-apiserver goes into crashloop backoff almost immediately and stays there, I'm not seeing a smoking gun but it seems like k8s isn't happy with the container runtime | 21:37:17 |
| sudoer | preflight isn't flagging any issues | 21:37:46 |
 Sheogorath | sudoer: what does your logs say? crictl should be able to provide you with them | 21:38:09 |
| sudoer | I see a lot of connection errors and timeouts to the local etcd server | 21:46:07 |
| sudoer | https://paste.centos.org/view/55ca9b2a | 21:49:24 |
| sudoer | Right after this, all the containers exited, crictl ps showed nothing running, now etcd is back up | 21:50:12 |
| sudoer | I don't see anything from etcd about it getting upset though | 21:50:54 |
| sudoer | https://paste.centos.org/view/a9d42590
That's etcd | 21:54:58 |
| Sheogorath | well, your etcd is shut down by a signal, which probably means its liveness probes fail (your kubelet logs might tell you that).
Further the apiserver fails because etcd isn't ready or listening on the specified backend (localhost). | 22:53:08 |
| 18 Apr 2024 |
 Karel | H | 05:44:24 |
 Rens Houben | I ? | 07:25:29 |
|  ramob joined the room. | 09:52:00 |
| DoofusCanadensis | J | 15:10:47 |
|  @rsaravanan:matrix.org joined the room. | 15:23:11 |
| @rsaravanan:matrix.org left the room. | 15:23:51 |
| sudoer | In reply to @sheogorath:shivering-isles.com
well, your etcd is shut down by a signal, which probably means its liveness probes fail (your kubelet logs might tell you that).
Further the apiserver fails because etcd isn't ready or listening on the specified backend (localhost). I needed to set the cgroup driver to systemd in the container runtime and kubelet | 16:20:47 |
| DoofusCanadensis | geez... in an AWS meeting asking about EKS and they're showing how we upgrade to kubernetes 1.32 | 17:56:14 |
 Maranda | 1.32...? 🤔 | 18:21:30 |
| * Maranda is like: wasn't 1.30 just released? | 18:23:48 |
 morguldir | $ aws --version aws-cli/1.32.34 Python/3.11.8 Linux/6.8.5-arch1-1 botocore/1.34.34 | 18:24:25 |
| morguldir | :D | 18:24:28 |
| Maranda | Or is Amazon forward in time with their elastic muppet | 18:25:50 |
| DoofusCanadensis | that was my comment to the presenter | 18:59:05 |
| DoofusCanadensis | guessing Amazon gets to work with very early releases | 18:59:25 |
 effendy | In reply to @sudoer:beeper.com
I needed to set the cgroup driver to systemd in the container runtime and kubelet I wanted to say that earlier, but I got completely distracted in the meantime :D Yes, this is the first thing to check, and it's really shitty, 'cause you cannot infer what's going on. | 19:01:49 |
| sudoer | Agreed, it'd be nice if preflight said something about it, it's not super obvious from the docs | 19:04:06 |
 jokeyrhyme | I just installed k3s on a cluster of 3x Raspberry Pi 4B nodes (8GB RAM)
The hardest part was figuring out my firewall rules, the k3s part was effortless :) | 23:06:54 |
| DoofusCanadensis | that it is | 23:12:09 |
| 19 Apr 2024 |
|  @spoonymac:matrix.org joined the room. | 06:02:56 |
