Upgrade flannel on kubernetes cluster

[michaelschefczyk]

Dear All,

In a debian buster kubernetes 1.20.1 cluster, I thought it might be a good idea to consider upgrading not only kubernetes but possibly also flannel. Hence, I ran

kubectl apply -f https://raw.githubusercontent.com/coreos/flannel/master/Documentation/kube-flannel.yml

once again. Now, I have both:

• quay.io/coreos/flannel v0.13.1-rc1 f03a23d55e57
• quay.io/coreos/flannel v0.12.0-amd64 4e9f801d2217

Both images are running and I have no clue (a) how to permanently stop one of the images and (b) if this would be on the right path do avioid duplications without creating addtional problems. I did look through all relevant daemonsets in all namespaces, for example. Nevertheless, I did not find anything to stop in order to really roll over to flannel v.0.13 (would have prepared to get stable instead of rc1 - but that would be another issue).

Can someone please point me to the right direction?

Thanks & Regards,

Michael Schefczyk

[michaelschefczyk]

For the time being, I did execute

kubectl delete daemonset kube-flannel-ds --namespace=kube-system

Then, only flannel 0.12.0 is running and I could delete the 0.13.1-rc1 image.

Nevertheless it would be good to know how to update, ideally to a stable version of choice.

[luthermonson]

needs documentation updates and probably will switch to a helm chart installation/uninstallation process.

[blachniet]

Any suggestions on the upgrade process here? Should we deploy kube-flannel-ds (v0.13.0) alongside the already running kube-flannel-ds-amd64 (v0.12.0), and then delete kube-flannel-ds-amd64? Is there a better approach?

[ckazimie]

Any update on how to correcty update flannel in an existing cluster?

[tarreislam]

Hi,

I did not find any information so i YOLO'D it on three production nodes, where 2 are clustered.

I will share my results for the next searcher

Host # 1, "the solo host" from: flannel:v0.13.1-rc1 to flannel:v0.15.1 Kubever: v1.20.2 OS: ubuntu server 18.04.03

Host #2 and #3, "the cluster" from: flannel:v0.11.1-rc1 to flannel:v0.15.1 Kubever: v1.20.5 OS: ubuntu server 18.04.06

The procedure for the solo node

1. kubectl delete daemonset kube-flannel-ds --namespace=kube-system
2. kubectl apply -f https://raw.githubusercontent.com/coreos/flannel/master/Documentation/kube-flannel.yml
3. Restart host
4. Hope for the best

The procedure for the clustered

1. kubectl drain slave-node
2. run steps from "solo node"-procedure but also deleting daemonsets without any pods
3. kubectl uncordon slave-node

And all my deployments, pods, services etc are working.

@ckazimie

[gusmenp]

Hi,

In the Documentation there is a file explaining the two procedures for upgrade/downgrade flannel https://github.com/flannel-io/flannel/blob/master/Documentation/upgrade.md

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.