Description of problem:
testcase "podman-system-service containers survive service stop" fail in 251-system-service.bats, this only found in cgroupv1 and runtime for crun.
When the system is cgroupv2, this case will success.
Version-Release number of selected component (if applicable):
podman-4.4.0-1.el9.x86_64
How reproducible:
100%
Steps to Reproduce:
1. set up cgroupv1 in RHEL9
2. check runtime is "crun"
3. bats 251-system-service.bats
Actual results:
[test@kvm-07-guest27 system]$ bats 251-system-service.bats
251-system-service.bats
✓ podman systerm service <bad_scheme_uri> returns error
✗ podman-system-service containers survive service stop
(from function `is' in file helpers.bash, line 689,
in test file 251-system-service.bats, line 50)
`is "$output" "true" "Container is still running after podman server stops"' failed
$ podman rm -t 0 --all --force --ignore
$ podman ps --all --external --format {{.ID}} {{.Names}}
$ podman images --all --format {{.Repository}}:{{.Tag}} {{.ID}}
quay.io/libpod/testimage:20221018 f5a99120db64
Running as unit: podman-service-SEC0EgR1iw.service
$ podman --url tcp://127.0.0.1:5245 run -d --name keeps-running quay.io/libpod/testimage:20221018 top -d 2
Resource limits are not supported and ignored on cgroups V1 rootless systems
7691f04bdd2d56af2daa61eba160a9d5e10ac85219b00ddbffb7da9aa56ce587
$ podman container inspect -l --format {{.State.Running}}
true
$ podman container inspect keeps-running --format {{.State.Running}}
false
#/vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
#| FAIL: Container is still running after podman server stops
#| expected: 'true'
#| actual: 'false'
#\^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
# [teardown]
$ podman pod rm -t 0 --all --force --ignore
$ podman rm -t 0 --all --force --ignore
7691f04bdd2d56af2daa61eba160a9d5e10ac85219b00ddbffb7da9aa56ce587
$ podman network prune --force
$ podman volume rm -a -f
✓ podman --host / -H options
3 tests, 1 failure
Expected results:
the testcase should success.
Additional info:
Comment 2Giuseppe Scrivano
2023-02-17 08:10:04 UTC
that is probably related to not creating a cgroup when running rootless on cgroupv1. Since anyway we cannot apply limits we don't even create one.
Before we invest time into this issue, how much do we care about this specific configuration (rootless + system service + cgroupv1) given the default on RHEL 9 is cgroupv2?
Comment 3Giuseppe Scrivano
2023-02-17 08:55:40 UTC
another problem: we will need to fix the OCI runtimes to create the scope and not try to configure it; so it is not only Podman but we need to change crun and runc as well. My vote is to skip the test and just add a warning to podman system service when it runs on rootless cgroupv1
Thanks for the input all. Giuseppe, please create a warning message upstream, and we'll close this out once it's merged. Yiqiao, please diable this test.
Comment 6Giuseppe Scrivano
2023-02-21 10:34:06 UTC
Description of problem: testcase "podman-system-service containers survive service stop" fail in 251-system-service.bats, this only found in cgroupv1 and runtime for crun. When the system is cgroupv2, this case will success. Version-Release number of selected component (if applicable): podman-4.4.0-1.el9.x86_64 How reproducible: 100% Steps to Reproduce: 1. set up cgroupv1 in RHEL9 2. check runtime is "crun" 3. bats 251-system-service.bats Actual results: [test@kvm-07-guest27 system]$ bats 251-system-service.bats 251-system-service.bats ✓ podman systerm service <bad_scheme_uri> returns error ✗ podman-system-service containers survive service stop (from function `is' in file helpers.bash, line 689, in test file 251-system-service.bats, line 50) `is "$output" "true" "Container is still running after podman server stops"' failed $ podman rm -t 0 --all --force --ignore $ podman ps --all --external --format {{.ID}} {{.Names}} $ podman images --all --format {{.Repository}}:{{.Tag}} {{.ID}} quay.io/libpod/testimage:20221018 f5a99120db64 Running as unit: podman-service-SEC0EgR1iw.service $ podman --url tcp://127.0.0.1:5245 run -d --name keeps-running quay.io/libpod/testimage:20221018 top -d 2 Resource limits are not supported and ignored on cgroups V1 rootless systems 7691f04bdd2d56af2daa61eba160a9d5e10ac85219b00ddbffb7da9aa56ce587 $ podman container inspect -l --format {{.State.Running}} true $ podman container inspect keeps-running --format {{.State.Running}} false #/vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv #| FAIL: Container is still running after podman server stops #| expected: 'true' #| actual: 'false' #\^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ # [teardown] $ podman pod rm -t 0 --all --force --ignore $ podman rm -t 0 --all --force --ignore 7691f04bdd2d56af2daa61eba160a9d5e10ac85219b00ddbffb7da9aa56ce587 $ podman network prune --force $ podman volume rm -a -f ✓ podman --host / -H options 3 tests, 1 failure Expected results: the testcase should success. Additional info: