This message was deleted.
# generala
adamant-kite-43734
04/18/2023, 10:25 AMThis message was deleted.
c
creamy-pencil-82913
04/18/2023, 5:07 PMcreamy-pencil-82913
04/18/2023, 5:08 PMOn all currently available releases, you need to add a extra mount for the log directory in order to get the log file actually written to the host
b
busy-flag-55906
04/19/2023, 9:11 AMthanks, will try this
busy-flag-55906
04/19/2023, 9:11 AMis there any full reference for those options? https://github.com/rancher/rke2/issues/1183#issuecomment-1242031356
busy-flag-55906
04/19/2023, 12:55 PMalso, i have noticed that 2 of 3 master nodes have problems with rke-server service
busy-flag-55906
04/19/2023, 12:58 PMand it seems that kube api manifests have been updated, but not applied to the pod
c
creamy-pencil-82913
04/19/2023, 2:25 PMI suspect you have a syntax error somewhere. What's the complete configuration file look like?
b
busy-flag-55906
04/20/2023, 7:35 AMhere is my current spec:
busy-flag-55906
04/20/2023, 7:36 AMthe only settings i add are :
c
creamy-pencil-82913
04/20/2023, 7:46 AMwhat does the resulting config on the rke2 node look like? under /etc/rancher/rke2/config.yaml.d ?
creamy-pencil-82913
04/20/2023, 7:46 AMI believe Rancher does some filtering of the config, you can’t set everything via the cluster editor. You might have to drop the required config directly on the node.
b
busy-flag-55906
04/20/2023, 8:57 AMcurrent configuration results in mounts with readonly fs for logs collection
busy-flag-55906
04/20/2023, 9:02 AMchanging audit output path to audit-log-path=- will produce another issues where 1 master is working fine and all logs are logged to stdout but 2 other masters remain in broken state
c
creamy-pencil-82913
04/20/2023, 3:41 PMCan you try putting the policy file somewhere other than where you’re going to be writing logs?audit-policy-file=/var/log/audit-k8s/policy-test.yaml
creamy-pencil-82913
04/20/2023, 3:42 PMThe log output mount shouldn’t have any other files mounted under it, and the policy file would violate that
b
busy-flag-55906
04/21/2023, 7:59 AMi have moved audit policy file to a different dir, however the issue with read only mount remains:
c
creamy-pencil-82913
04/21/2023, 3:15 PMDo you have selinux policy or something else that is blocking the write? We don't set up the mount as read-only.
creamy-pencil-82913
04/21/2023, 3:15 PMIt is just a normal hostpath mount
b
busy-flag-55906
04/24/2023, 7:07 AMno, i do not have anything like this, inside kube-api manifest this path is set as read-only once i add it
busy-flag-55906
04/24/2023, 9:25 AMthe things become more weirder, finally the same configuration in cluster have been updated on all masters, but another issue is that 2 or 3 servers have kube-api running with incorrect settings, in manifest i see the following:
busy-flag-55906
04/24/2023, 9:27 AMcrictl inspect $POD :
busy-flag-55906
04/24/2023, 9:28 AM/etc/rancher/rke2/config.yaml.d/50-rancher.yaml:
busy-flag-55906
04/24/2023, 9:37 AMi have just noticed that kubelet is not able to start properly, probably thats why kube-api havent restarted yet
8 Views