chore(deps): update dependency containers/automation_images to v20240529

[renovate]

This PR contains the following updates:

Package	Update	Change
containers/automation_images	major	20240513t140131z-f40f39d13 -> 20240529t141726z-f40f39d13

---

Release Notes

containers/automation_images (containers/automation_images)

v20240529t141726z-f40f39d13

Compare Source

---

Configuration

📅 Schedule: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined).

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR is behind base branch, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR has been generated by Mend Renovate. View repository job log here.

[edsantiago]

I think the failure is (log):

ValidateConfig [It] should succeed with default config
/var/tmp/go/src/github.com/containers/common/pkg/config/config_test.go:21
  [FAILED] Expected
      <string>: zstd:chunked
  to be equivalent to
      <string>: gzip
  In [It] at: /var/tmp/go/src/github.com/containers/common/pkg/config/config_test.go:35

Not something I can fix.

[rhatdan]

This means that the VMs are using the containers.conf from Rawhide?

[rhatdan]

We probably should change the default in containers.conf to zstd:chunked and then change containers.conf for RHEL and Fedora < 41 to default to gzip.

[rhatdan]

@giuseppe @Luap99 @mheon @baude WDYT?

[Luap99]

This count as breaking change to me. The upstream defaults will be consumed by all distros so if we change this in a minor release we could make many people unhappy if it breaks for whatever reasons.

Having a documented process to turn it on for the next fedora release (https://fedoraproject.org/wiki/Changes/zstd:chunked) is fine but forcing this into all distros maybe not so much. Looking at containers/storage#1928 there seem to be cases where it performs much worse currently.

[Luap99]

Regardless of this specific case unit tests should never depend on installed system configs at all. So the proper fix is to make the unit tests not read system configs and use them to check defaults as this is broken no matter what and true for every config field we have

[giuseppe]

Regardless of this specific case unit tests should never depend on installed system configs at all. So the proper fix is to make the unit tests not read system configs and use them to check defaults as this is broken no matter what and true for every config field we have

I agree, there is no point in what we are testing now.

Should we just use containers.conf to read the configuration?

diff --git a/pkg/config/config_test.go b/pkg/config/config_test.go
index bacab043..f48c085e 100644
--- a/pkg/config/config_test.go
+++ b/pkg/config/config_test.go
@@ -21,7 +21,7 @@ var _ = Describe("Config", func() {
                It("should succeed with default config", func() {
                        // Given
                        // When
-                       defaultConfig, err := NewConfig("")
+                       defaultConfig, err := NewConfig("containers.conf")

[Luap99]

Regardless of this specific case unit tests should never depend on installed system configs at all. So the proper fix is to make the unit tests not read system configs and use them to check defaults as this is broken no matter what and true for every config field we have

NewConfig() still reads all system configs by default + the given path so it doesn't help us here unfortunately.
We could switch directly to defaultConfig() and readConfigFromFile() but then we also loose coverage from the existing walk all paths logic + the default validation. Maybe the easiest fix is to set CONTAINERS_CONF env as this one will skip all system configs. We loose coverage for this but it should not be a big deal as podman/buildah integrations tests can still catch issues

[rhatdan]

If we tested "/dev/null" we would get the default config

[cevich]

...deleted...

[openshift-ci]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: cevich, renovate[bot]
Once this PR has been reviewed and has the lgtm label, please ask for approval from luap99. For more information see the Kubernetes Code Review Process.

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

Needs approval from an approver in each of these files:

• OWNERS

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

[cevich]

ISTM like something might just need updating in the tests:

------------------------------
• [FAILED] [0.002 seconds]
Config 
ValidateConfig 
[It] should succeed with default config
/var/tmp/go/src/github.com/containers/common/pkg/config/config_test.go:21
[FAILED] Expected
      <string>: zstd:chunked
  to be equivalent to
      <string>: gzip
In 
[It]
 at: 
/var/tmp/go/src/github.com/containers/common/pkg/config/config_test.go:35
@ 06/07/24 16:19:53.26
------------------------------

[cevich]

@vrothberg @Luap99 @rhatdan any hope of fixing the problem here? c/common is a pretty low-level dep. It would be nice to have it using the latest CI VM images everything else is already using.

[Luap99]

see comments above for how to fix.

In the short time we get a working f40 containers-common again, ref #2048,
so rebuilding newer images with the correct containers-common might be the fastest fix.
Or well just patch the config file in our CI setup here if you really want the new image merged.

[cevich]

so rebuilding newer images with the correct containers-common might be the fastest fix.

Oh, that I know how to do.

Or well just patch the config file in our CI setup here

I'm no expert, but skimming the comments above, it sounds like that might also be desirable?

[cevich]

Force-push: Jammed in a recent image set from containers/automation_images#354 (comment)

[cevich]

Ugh, since renovate just overwrote my commit w/ newer images, they do indeed pass the F40 test.

[cevich]

...deleted: wrong tab...

[cevich]

#2059 will fix this.