WIP: Cirrus: Add Ubuntu to rootless integration testing

[cevich]

Fixes #9508

Signed-off-by: Chris Evich [email protected]

[rhatdan]

LGTM
I like more tests.

[rhatdan]

/approve

[openshift-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: cevich, rhatdan

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [rhatdan]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[rhatdan]

Sadly test are breaking.

[cevich]

Sadly test are breaking.

That's a matter of perspective; One could say the tests have freshly uncovered previously hidden problems 😁

• int podman ubuntu-2004 rootless host
• int podman ubuntu-2010 rootless host

Let's see what we got going on here...

[cevich]

...okay, these appear like fairly clear problems:

• For Ubuntu 20.10 (crun), only one test failed with a pretty clear error: error starting container a7a...6c2: the requested cgroup controller cpu is not available: OCI runtime error. Assuming that means some thing like "The kernel is too old", maybe this is a test we can simply skip on Ubuntu 20.10?
• For Ubuntu 20.04 (runc), all the errors are capability related, so again perhaps a "kernel is too old" issue. @rhatdan what do we do in case of Error: OCI runtime error: container_linux.go:367: starting container process caused: unknown capability "CAP_PERFMON"?

[Luap99]

The Error: OCI runtime error: container_linux.go:367: starting container process caused: unknown capability "CAP_PERFMON" error only happens in network tests. I completely rework rootless networking in #9423. Maybe it will work with the new networking logic. I added this commit to my PR lets see if it will work there.

[cevich]

@Luap99 thanks! I also have two PRs to update the VM images, both are stuck waiting for an updated crun, which could (maybe) have some bearing here, esp. on the first (Ubuntu 20.10) problem.

[Luap99]

Ubuntu 20.04 works with my PR https://github.com/containers/podman/pull/9423/checks?check_run_id=2238965171.

[edsantiago]

New map:

[cevich]

@Luap99 excellent news, does it also pass standalone and/or do you want me to rebase this?

@giuseppe or @rhatdan any ideas on the error starting container a7a...6c2: the requested cgroup controller 'cpu' is not available: OCI runtime error?

[Luap99]

@cevich I think the fix for the cpu cgroup controller failure is just to skip the test, see:

podman/test/e2e/play_kube_test.go

Lines 1830 to 1833 in 5798102

	It("podman play kube allows setting resource limits", func() {
	SkipIfContainerized("Resource limits require a running systemd")
	SkipIfRootlessCgroupsV1("Limits require root or cgroups v2")
	SkipIfUnprivilegedCPULimits()

However the SkipIfUnprivilegedCPULimits only skips when rootless on fedora

podman/test/e2e/common_test.go

Lines 608 to 613 in 5798102

	func SkipIfUnprivilegedCPULimits() {
	info := GetHostDistributionInfo()
	if isRootless() && info.Distribution == "fedora" {
	ginkgo.Skip("Rootless Fedora doesn't have permission to set CPU limits")
	}
	}

So I guess we have to add ubuntu 20.10 there as well.

[cevich]

So I guess we have to add ubuntu 20.10 there as well.

Ahh perfect. @giuseppe or @rhatdan un-ping, and sorry for the noise. Let's see if we can just remove && info.Distribution == "fedora"...

[cevich]

Rebased onto #9423 then again onto master.

[cevich]

Rebased on master

[cevich]

Rebased on master

[cevich]

Blocked by #9997

[cevich]

Rebased on master

[cevich]

Obsolete by #10237

[openshift-ci]

@cevich: PR needs rebase.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.