v4.4.1-rhel-frankenstein

[lsm5]

Checklist

Ensure you have completed the following checklist for your pull request to be reviewed:

• Certify you wrote the patch or otherwise have the right to pass it on as an open-source patch by signing all
  commits. (git commit -s). (If needed, use git commit -s --amend). The author email must match
  the sign-off email address. See CONTRIBUTING.md
  for more information.
• Referenced issues using Fixes: #00000 in commit message (if applicable)
• Tests have been added/updated (or no tests are needed)
• Documentation has been updated (or no documentation changes are needed)
• All commits pass make validatepr (format/lint checks)
• Release note entered in the section below (or None if no user-facing changes)

Does this PR introduce a user-facing change?

TBD

[cevich]

Testing of 044d71e is still ongoing, since I'm also running the "remote" tests this time. The local results are ready though.

044d71e-local_rootless_system.txt
044d71e-local_root_system.txt

By my 👁️ these appear unchanged compared to prior runs. @lsm5 since you're fixing system tests, any thoughts on patching the string/exit code mismatches resulting from running w/ runc? Specifically these two I'm seeing over and over:

not ok 39 podman run - basic tests
# (from function `is' in file ./helpers.bash, line 706,
#  in test file ./030-run.bats, line 39)
#   `is "$output" "$expected_output" "podman run $cmd - output"' failed
...cut...
# $ podman run quay.io/libpod/testimage:20221018 /etc
# Error: OCI runtime error: runc: runc create failed: unable to start container process: error during container init: exec: "/etc": is a directory
# [ rc=126 (expected) ]
# #/vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
# #|     FAIL: podman run /etc - output
# #| expected: 'Error:.*exec.*permission denied' (using expr)
# #|   actual: 'Error: OCI runtime error: runc: runc create failed: unable to start container process: error during container init: exec: "/etc": is a directory'
# #\^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
#

and

not ok 201 podman exec - basic test
# (from function `die' in file ./helpers.bash, line 558,
#  from function `run_podman' in file ./helpers.bash, line 228,
#  in test file ./075-exec.bats, line 25)
#   `run_podman 126 exec $cid /etc' failed
...cut...
# $ podman exec 70b3f9c02e132569a91c0641415b548f5dc298315bc7caf92cf22413b57bddce /etc
# Error: OCI runtime error: runc: exec failed: unable to start container process: exec: "/etc": is a directory
# [ rc=255 (** EXPECTED 126 **) ]
# #/vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
# #| FAIL: exit code is 255; expected 126
# #\^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
#

Though perhaps the fact that one exits 126 and the other 125 is also unexpected behavior?

[cevich]

Remote results (nothing new/exciting), basically the same as the "local" tests by my 👁️

044d71e-remote_rootless_system.txt
044d71e-remote_root_system.txt

[lsm5]

@cevich some more updates. PTAL at the latest.

[lsm5]

@cevich some more updates. PTAL at the latest.

The run and exec failures in rootful should be fixed with the latest. I have yet to run them as rootless, but please give it a try.

[cevich]

RHEL 9.2 system test results:

• The local rootless tests are all passing now. The only exception are pasta tests that require IPv6 which is non-functional/unavailable in this environment.
• The local root tests had two failures. One was IPv6 related so can be ignored. The other I think it's related to the "test/system: adjust pod memory limit for runc minimum requirement" commit changes:

not ok 305 pod resource limits
# (from function `die' in file ./helpers.bash, line 558,
#  from function `run_podman' in file ./helpers.bash, line 228,
#  in test file ./200-pod.bats, line 528)
#   `run_podman --cgroup-manager=$cgm pod start $name' failed
# # /usr/bin/podman rm -t 0 --all --force --ignore
# # /usr/bin/podman ps --all --external --format {{.ID}} {{.Names}}
# # /usr/bin/podman images --all --format {{.Repository}}:{{.Tag}} {{.ID}}
# quay.io/libpod/testimage:20221018 f5a99120db64
# # /usr/bin/podman --cgroup-manager=systemd pod create --name=resources-systemd --cpus=5 --memory=6m --memory-swa
p=1g --cpu-shares=1000 --cpuset-cpus=0 --cpuset-mems=0 --device-read-bps=/dev/loop0:1mb --device-write-bps=/dev/lo
op0:1mb --blkio-weight=50
# 551c5819d262c87fdf0914b1803391b658e6b32acc6631ff697beec2e764530e
# # /usr/bin/podman --cgroup-manager=systemd pod start resources-systemd
# /usr/lib/bats-core/test_functions.bash: line 254: warning: command substitution: ignored null byte in input
# Error: starting container d01c900ad0a41cc47798fc9f42021bdfa8502721618f81a2d57d921db40e5c67: container create fai
led (no logs from conmon): conmon bytes "": readObjectStart: expect { or n, but found , error found in #0 byte of
...||..., bigger context ...||...
# [ rc=125 (** EXPECTED 0 **) ]

• remote root also had a single failure in podman-remote: non-default connection
  fedda4a_remote_root.txt
  This is the same as I saw before, where it seems the test is expecting a default connection
  to be absent. I haven't reviewed the test code, but on the surface that logic seems flawed.
• remote rootless was nothing but the expected ip6 pasta failures

[lsm5]

• The local root tests had two failures. One was IPv6 related so can be ignored. The other I think it's related to the "test/system: adjust pod memory limit for runc minimum requirement" commit changes:

Fixed

• remote root also had a single failure in podman-remote: non-default connection
  fedda4a_remote_root.txt

This seems like a flake as it passed on manual run on testing-farm instance. Pushed another commit to unset PODMAN_SOCKET . Hopefully it passes for you too 🤞

[cevich]

Starting another run vs 1720520

[cevich]

Results:

• 1720520_local_rootless_system.txt ✔️ Only the expected ipv6/pasta problems we can ignore*.
• 1720520_local_root_system.txt ✔️
• 1720520_remote_rootless_system.txt ✔️ Same as local_rootless*
• 1720520_remote_root_system.txt ✔️

---

*Note: I just noticed one of the pasta failures doesn't explicitly/directly indicate in its output anything about local IPv6 unavailability. I was assuming it was related, but perhaps not?

not ok 505 podman networking with pasta(1) - ICMPv6 echo request
# (from function `die' in file ./helpers.bash, line 558,
#  from function `run_podman' in file ./helpers.bash, line 228,
#  in test file ./505-networking-pasta.bats, line 672)
#   `run_podman run --net=pasta $IMAGE ping -6 -c3 -W1 \' failed
# $ podman rm -t 0 --all --force --ignore
# 0419269548ecd393edf55500b0eba715908b2356e272037457acb9ce19a3a73c
# $ podman ps --all --external --format {{.ID}} {{.Names}}
# $ podman images --all --format {{.Repository}}:{{.Tag}} {{.ID}}
# quay.io/libpod/testimage:20221018 f5a99120db64
# $ podman run --net=pasta quay.io/libpod/testimage:20221018 ping -6 -c3 -W1 sh -c ping
 -c3 -W1 sed -nr "s/^nameserver[ ]{1,}([^:]*):(.*)/\1:\2/p" /etc/resolv.conf | head -1
# ping: invalid number 'ping -c3 -W1 sed -nr "s/^nameserver[ ]{1,}([^:]*):(.*)/\1:\2/p"
 /etc/resolv.conf | head -1'
# [ rc=1 (** EXPECTED 0 **) ]

[cevich]

Looking at 505-networking-pasta.bats, line 672, this makes no sense to me at all. Neither does the IPv4 version of the test 😕

Confirmed, I ran the test commands manually -- both the podman networking with pasta(1) - ICMP echo request are broken.

1. The IPv4 version results in a busybox usage message as output, but then it exits 0. So BATS is treating this as a passing test.
2. The IPv4 version fails with exactly the output as seen from BATS above: ping: invalid number.

I believe this is what was intended (I tried the IPv4 one and it works):

    run_podman run --net=pasta $IMAGE \
        sh -c 'ping -c3 -W1 $(sed -nr "s/^nameserver[ ]{1,}([^.]*).(.*)/\1.\2/p" /etc/resolv.conf | head -1)'

and

    run_podman run --net=pasta $IMAGE \
        sh -c 'ping -6 -c3 -W1 $(sed -nr "s/^nameserver[ ]{1,}([^:]*):(.*)/\1:\2/p" /etc/resolv.conf | head -1)'