CI fails on almalinux 9

[kolyshkin]

Almalinux-9 CI job (which runs on Cirrus CI) is consistently timing out for the last week or so.

I took a look, and it seems it's caused by a recent kernel update:

• was working file with 5.14.0-427.33.1.el9_4.x86_64;
• started failing (timing out) with 5.14.0-427.37.1.el9_4.x86_64.

The CI job times out while executing the "update paused container" integration test, here it is:

https://github.com/opencontainers/runc/blob/d82235c9d0e249c89f136811b97c9277eae80f8c/tests/integration/update.bats#L830C1-L853C1

I have seen it being timed out on both systemd and fs driver, and on both root and rootless tests.

[kolyshkin]

Unable to reproduce locally, but disabling the test for almalinux-9 apparently fixes the issue.

[kolyshkin]

The CI job times out while executing the "update paused container" integration test

I was wrong, this is probably the next test, i.e. "update memory vs CheckBeforeUpdate". Here it is:

runc/tests/integration/update.bats

Lines 854 to 892 in d82235c

	@test "update memory vs CheckBeforeUpdate" {
	requires cgroups_v2
	[ $EUID -ne 0 ] && requires rootless_cgroup
	runc run -d --console-socket "$CONSOLE_SOCKET" test_update
	[ "$status" -eq 0 ]
	# Setting memory to low value with checkBeforeUpdate=true should fail.
	runc update -r - test_update <<EOF
	{
	"memory": {
	"limit": 1024,
	"checkBeforeUpdate": true
	}
	}
	EOF
	[ "$status" -ne 0 ]
	[[ "$output" == *"rejecting memory limit"* ]]
	testcontainer test_update running
	# Setting memory+swap to low value with checkBeforeUpdate=true should fail.
	runc update -r - test_update <<EOF
	{
	"memory": {
	"limit": 1024,
	"swap": 2048,
	"checkBeforeUpdate": true
	}
	}
	EOF
	[ "$status" -ne 0 ]
	[[ "$output" == *"rejecting memory+swap limit"* ]]
	testcontainer test_update running
	# The container will be OOM killed, and runc might either succeed
	# or fail depending on the timing, so we don't check its exit code.
	runc update test_update --memory 1024
	wait_for_container 10 1 test_update stopped
	}