When a users asks for specific devices we should still add them and not
ignore them just because privileged adds all of them.
Most notably if you set --device /dev/null:/dev/test you expect
/dev/test in the container, however as we ignored them this was not the
case. Another side effect is that the input was not validated at at all.
This leads to confusion as descriped in the issue.
Fixes#23132
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
I found that Quadlet didn't currently have support for log options.
This merge allows Quadlet to handle log options and correctly
pass those values through to `podman run` for Container and Kube
types.
Syntactically consistent with existing parameters:
```ini
[Container]
Image=localhost/imagename
LogOpt=path=/var/log/container/mycontainer.json
LogOpt=size=10mb
```
Signed-off-by: Brett Calliss <brett@obligatory.email>
When we execute ps(1) in the container and the container uses a userns
with a different id mapping the user id field will be wrong.
To fix this we must join the userns in such case.
Fixes#22293
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Two tests are skipped for a long time because they flaked to much,
nobody cares about them and there are only debugging endpoints mostly so
it is not critical either.
The "of 2 seconds" tests isn't useful either. It waits up to 30s for the
exit so it doesn't actually verify a proper timeout. Additionally we
have similar checks in the system tests "podman system service -
CORS enabled in logs" so I consider this safe to remove.
Fixes#12624
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
The pod was set after we checked the namespace and the namespace code
only checked the --pod flag but didn't consider --pod-id-file option.
As such fix the check to first set the pod option on the spec then use
that for the namespace. Also make sure we always use an empty default
otherwise it would be impossible in the backend to know if a user
requested a specific userns or not, i.e. even in case of a set
PODMAN_USERNS env a container should still get the userns from the pod
and not use the var in this case. Therefore unset it from the default
cli value.
There are more issues here around --pod-id-file and cli validation that
does not consider the option as conflicting with --userns like --pod
does but I decided to fix the bug at hand and don't try to fix the
entire mess which most likely would take days.
Fixes#22931
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Do not return 200 status code before we know if there will be an error.
Delay writing the status code until we send the first response. That way
we can set an error code inside the loop when we get a error on the
first try, i.e. because an invalid descriptor was used.
Fixes#22986
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
When we failed to do anything we should return 500, the 409 code has a
special meaing to the client as it uses a different error format. As
such the remote client was not able to unmarshal the error correctly and
just returned an empty string.
Fixes#22989
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Add a test to verify that restoring a container in a Pod works when
the `container restore --pod` option is used with Pod *name* (this
functionality was previously limited to support only full Pod ID).
Signed-off-by: Radostin Stoyanov <rstoyanov@fedoraproject.org>
This is the same as what --squash-all is doing, and we already support
--squash with --layers=true since this is the default.
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
If a container was stopped and we try to start it before we called
cleanup it tried to reuse the network which caused a panic as the pasta
code cannot deal with that. It is also never correct as the netns must
be created by the runtime in case of custom user namespaces used. As
such the proper thing is to clean the netns up first.
Also change a e2e test to report better errors. It is not directly
related to this chnage but it failed on v1 of this patch so we noticed
the ugly error message it produced. Thanks to Ed for the fix.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
debian's man (5) hostname page states "The file should contain a single newline-terminated hostname
string."
[NO NEW TESTS NEEDED]
fix#22729
Signed-off-by: Bo Wang <wangbob@uniontech.com>
The e2e tests already depend on skopeo anyway and pulling a over 300
MB image is not helpful for flakes but most importantly we see ENOSPC
flakes. I see them around the skopeo test so I assume the big image is
pushing the tmpfs limits so other tests running in parallel can start
failing because of it.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
the condition doesn't work when the runtime to use is specified
through its absolute path as the error message contains that.
Simplify the check and just look for "read from the init process".
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
As agreed in Planning meeting of 2024-03-20, Podman 5.x will
drop support for cgroups v1 and for runc. Make it so.
CI images built in https://github.com/containers/automation_images/pull/338
Signed-off-by: Ed Santiago <santiago@redhat.com>
.build files allow to build an image via Quadlet. The keys from a .build
file are translated to arguments of a `podman build` command by Quadlet.
Minimal keys for .build files are `ImageTag=` and a context directory,
see `SetWorkingDirectory=`, or a `File=` pointing to a Containerfile.
After sorting .build files into the Quadlet dependency order, there
remains a possible dependency cycle issue between .volume and .build
files: A .volume can have `Image=some.build`, and a .build can have
`Volume=some.volume:/some/volume`.
We solve this dependency cycle by prefilling resourceNames with all
image names from .build files before converting all the unit files.
This results in an issue for the test suite though: For .volume's
depending on *.image or *.build, we need to copy these additional
dependencies to the test's quadletDir, otherwise the test will fail.
This is necessary, because `handleImageSource()` actually needs to know
the image name defined in the referenced *.{build,image} file. It cannot
fall back on the default names, as it is done for networks or volumes,
for example.
Signed-off-by: Johannes Maibaum <jmaibaum@gmail.com>
The new c/image version is returning a slightly new error message[1] so
make tests use the new one.
[1] https://github.com/containers/image/pull/2408
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
When an empty volume is mounted into a container, Docker will
chown that volume appropriately for use in the container. Podman
does this as well, but there are differences in the details. In
Podman, a chown is presently a one-and-done deal; in Docker, it
will continue so long as the volume remains empty. Mount into a
dozen containers, but never add content, the chown occurs every
time. The chown is also linked to copy-up; it will always occur
when a copy-up occurred, despite the volume now not being empty.
This PR changes our logic to (mostly) match Docker's.
For some reason, the chowning also stops if the volume is chowned
to root at any point. This feels like a Docker bug, but as they
say, bug for bug compatible.
In retrospect, using bools for NeedsChown and NeedsCopyUp was a
mistake. Docker isn't actually tracking this stuff; they're just
doing a copy-up and permissions change unconditionally as long as
the volume is empty. They also have the two linked as one
operation, seemingly, despite happening at very different times
during container init. Replicating that in our stateful system is
nontrivial, hence the need for the new CopiedUp field. Basically,
we never want to chown a volume with contents in it, except if
that data is a result of a copy-up that resulted from mounting
into the current container. Tracking who did the copy-up is the
easiest way to do this.
Fixes#22571
Signed-off-by: Matthew Heon <matthew.heon@pm.me>
If a container unit starts on boot with a dependency on `default.target`
the image unit may start too soon, before network is ready. This cause
the unit to fail to pull the image.
- Add a dependency on `network-online.target` to make sure image pulls
don't fail.
See https://github.com/containers/podman/issues/21873
- Document the hardcoded dependency on `network-online.target` for images unit
and explain how it can be overriden if necessary.
- tests/e2e/quadlet: Add `assert-last-key-regex`
Required to test the `After=` override in [Unit] section
See https://github.com/containers/podman/pull/22057#issuecomment-2008959993
- quadlet/unitfile: add a prepenUnitLine method
Requirements on networks should be inserted at the top of the
section so the user can override them.
Signed-off-by: jbtrystram <jbtrystram@redhat.com>
Final followup to #22270. That PR added a temporary convention
allowing a new form of ExitWithError(), one with an exit code
and stderr substring. In order to allow bite-size progress,
the old no-args form was still allowed. This PR removes
support for no-args ExitWithError().
This PR also adds one piece of new functionality: passing ""
(empty string) as the stderr arg means "expect exit code
but fail if there's anything at all in stderr".
Signed-off-by: Ed Santiago <santiago@redhat.com>
Follow up to commit eaf60c7fe7, with the toolbox image removal it is
possible to run all tests from tmpfs.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Followup to #22270: wherever possible/practical, extend command
error checks to include explicit exit status codes and error strings.
This commit handles only one file, test/e2e/rmi_test.go , because
my changes are significant enough to merit individual review.
Signed-off-by: Ed Santiago <santiago@redhat.com>
