flatcar-linux-update-operator

flatcar-linux-update-operator copied to clipboard

Expose prometheus metrics from agent and operator

6

Original issues: - https://github.com/coreos/container-linux-update-operator/issues/157 - https://github.com/coreos/container-linux-update-operator/issues/156 Do note, that some of this information can already be obtained using `kube_node_labels` metric: ``` kube_node_labels{container="kube-state-metrics",endpoint="http",job="kube-state-metrics",label_beta_kubernetes_io_arch="amd64",label_beta_kubernetes_io_os="linux",label_flatcar_linux_update_v1_flatcar_linux_net_group="alpha",label_flatcar_linux_update_v1_flatcar_linux_net_id="flatcar",label_flatcar_linux_update_v1_flatcar_linux_net_reboot_needed="false",label_flatcar_linux_update_v1_flatcar_linux_net_version="2671.0.0",label_foo="bar",label_kubernetes_io_arch="amd64",label_kubernetes_io_hostname="theory-main-workers-1",label_kubernetes_io_os="linux",namespace="prometheus-operator",node="theory-main-workers-1",pod="prometheus-operator-kube-state-metrics-d969c4b58-68zqw",service="prometheus-operator-kube-state-metrics"}

invidian

WIP: pkg/agent: wait for all volumes to be detached before rebooting

3

This commit provides PoC version of implementing agent waiting for all volumtes attached to the node to be detached as a step after draining the node, as shutting down the...

invidian

Ensure update-agent waits for all volumes to be detached before rebooting

8

Original issue: https://github.com/coreos/container-linux-update-operator/issues/191 Perhaps waiting for `kubectl get volumeattachments` to get empty with the right selector would be sufficient?

invidian

[RFE] Add delay between reboots

4

## Current situation Hi, We have a K8ssandra cluster running on our K8s cluster. Flatcar reboots quite quickly but some applications might take longer to startup and initialize. The Flatcar...

mikekuzak

We have e.g. https://github.com/flatcar-linux/flatcar-linux-update-operator/blob/53f08043e320c853940ed7b4c126c7b72af1af00/examples/reboot-annotations/after-reboot-daemonset.yaml right now, but it only annotates the node right away, which is not very useful, as it does not allow implementing any logic. We should change...

invidian

Add unit tests with fake client-go for all parts of controller

2

Given that this operator simply operates on Node objects, we should be able to use client-go to run both operator and node agent usign fake client-go and with mock D-Bus...

invidian

Ensure update-agent does not panics/segfaults when node is shutting down

1

Original issue: https://github.com/coreos/container-linux-update-operator/issues/184 This has been reported in the past and partially resolved, but it seems it is not finished. See also coreos/container-linux-update-operator#184 We should make sure this is addressed,...

invidian

update agent fails to watch for ok-to-reboot label

1

With following error: ``` W0206 12:37:02.081765 1 agent.go:152] error waiting for an ok-to-reboot: waiting for annotation "flatcar-linux-update.v1.flatcar-linux.net/reboot-ok" failed: unhandled watch case for watch.Event{Type:"ERROR", Object:(*v1.Status)(0xc00024abe0)} ```

invidian

That make no sense and only pollutes the logs. This however, may require us to start testing log messages in the agent, which is currently not done, but definitely desirable.

invidian

Fix remaining existing linter issues

1

There are still some which should be fixed: ``` $ golangci-lint run | grep \\.go: | wc -l 64 ```

invidian