ilia1243
ilia1243
`kubemarine check_paas --tasks kubernetes.pods` does not show pods in status `CrashLoopBackOff` if their containers fail to start.
We `kubectl uncordon` with `warn=True` https://github.com/Netcracker/KubeMarine/blob/v0.29.0/kubemarine/system.py#L367 ```python res = first_control_plane.sudo(f'kubectl uncordon {node_name}', warn=True) ``` This may leave nodes unschedulable that is especially relevant for All-in-One scheme. Potential remediation is to...
### Description * If some kubernetes image version downgrades in the compatibility map, it may still not downgrade during Kubernetes upgrade. * This leads to inconsistency of expected and actual...
### Description * Since the upstream kubeadm issue is resolved and will be probably included in Kubernetes v1.31, plan to remove the work around. ### Regression testing Run Test Cases...
Restart of containers for static pods happens through `crictl rm` https://github.com/Netcracker/KubeMarine/blob/v0.28.0/kubemarine/kubernetes/components.py#L999 that is not idiomatic according to https://github.com/kubernetes/kubernetes/issues/123177 Need to think if it is possible to delete & create the...
Need to elaborate the purpose of https://github.com/Netcracker/KubeMarine/blob/v0.27.0/kubemarine/resources/configurations/compatibility/kubernetes_versions.yaml#L2 and probably add this option to the `kubemarine config` procedure coming soon https://github.com/Netcracker/KubeMarine/pull/613#discussion_r1511133305
Based on the thread https://github.com/Netcracker/KubeMarine/pull/473#discussion_r1269421495 it is necessary to assess and implement: 1. Support GPG-keys for `services.packages.package_manager.repositories` on RHEL-like OS families. 2. Install necessary public repositories by default in public...
After #592 is merged, it makes sense to implement some more cluster reconfiguration scenarios: 1. Remove `deploy.kubernetes.audit` installation task, and support to reconfigure `services.audit.cluster_policy` instead. 2. Support to reconfigure `cluster_name`.
Currently workflows are run on push event that restricts external contribution from forks. Need to run workflows on pull_request event and think of * Avoid double runs * Persist ability...
Consider implementing new dry-run or replace existing --without-act. * Do not change anything on cluster or in input Kubemarine files. * Generate, enrich, compile as much as possible * Compile...