RFE: backup manifest dir gets wiped after upgrade

[jipperinbham]

What keywords did you search in kubeadm issues before filing this one?

• temp
• tmp

BUG REPORT

Versions

kubeadm version (use kubeadm version): 1.8.0

Environment:

• Kubernetes version (use kubectl version): 1.7.6
• Cloud provider or hardware configuration: aws

What happened?

While watching TGIK, @jbeda ran sudo kubeadm upgrade apply v1.8.0 and the upgrade ran successfully but the previous manifests in /etc/kubernetes/tmp were deleted.

What you expected to happen?

The general comments were the previous manifests should have been kept around even with a successful upgrade.

How to reproduce it (as minimally and precisely as possible)?

• sudo kubeadm upgrade apply v1.8.0
• ls -alh /etc/kubernetes/tmp

Anything else we need to know?

This looks to be the expected behavior per https://github.com/kubernetes/kubernetes/blob/73d1b38604d6f39ec77a94f0f411a46c765d1c4f/cmd/kubeadm/app/phases/upgrade/staticpods.go#L172 so the question is whether the current behavior needs to be altered.

[jbeda]

Thanks for filing the issue! It was on my list to follow up.

@luxas -- I think you wrote this code. What do you think about leaving the old manifests there so that users can see what changed. It also enables a (dangerous) rollback path.

[luxas]

If you think the command should "leak" directories for the user -- feel free to change.
I thought that if the user actually wants to do that -- he/she would just do cp -r /etc/kubernetes{,-preupgrade} before kubeadm upgrade
Documentation improvement or code change?

[jipperinbham]

@luxas @jbeda I submitted kubernetes/kubernetes#57116 as an attempt to address the concern here. If y'all think it's better to go the route of adding verbiage to the upgrade docs, I'm happy to go that route instead.

[luxas]

@jipperinbham Thanks! I think more documentation would be better 👍

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[timothysc]

/assign @liztio