This message was deleted.
# k3sa
adamant-kite-43734
04/30/2024, 2:12 PMThis message was deleted.
c
creamy-pencil-82913
04/30/2024, 4:11 PMno. pods are groups of processes. When the processes exit as part of the reboot, the pods go away and must be recreated. This is why it is usually recommended to use DaemonSet/Deployment/ReplicaSet to manage creation of pods, instead of just creating pods directly.
creamy-pencil-82913
04/30/2024, 4:12 PMthis is just how Kubernetes works
c
colossal-elephant-30024
04/30/2024, 6:23 PMThe pods are created through deployments. What I observe is that k3s will recreate itself after the host is rebooted, which means every reboot will start a new cluster instead of picking up previous state
colossal-elephant-30024
04/30/2024, 6:23 PMIs this something expected for k3s or there are some configs to change the behavior
c
creamy-pencil-82913
04/30/2024, 6:26 PMthat is not correct
creamy-pencil-82913
04/30/2024, 6:27 PMare you running this on a host where /var/lib/rancher is tmpfs, immutable, or otherwise discarded on reboot?
creamy-pencil-82913
04/30/2024, 6:27 PMthe only way it would be recreating a new cluster every time you reboot is if you have something that is wiping the paths on disk where cluster data is stored.
c
colossal-elephant-30024
04/30/2024, 6:30 PMwhere are the paths for the data store? I am using a azure rhel vm instance, so I don't think it has those constraints
c
creamy-pencil-82913
04/30/2024, 11:18 PMThe path I just said up above.
👍 1
2 Views