Since commit f250560a8043 the play kube command uses its own network.
this is racy be design because we create the network followed by
creating/running pod/containers. This means in the meantime another
prune or reset process could wipe out the network config because we have
to share the network config directory by design in the test.
The problem is we only have one host netns which is shared between
tests. If the network config dir is not shared we cannot make conflict
checks for interface names and ip address. This results in different
tests trying to use the same interface and/or ip address which will
cause runtime failures in CNI and netavark.
The only solution I see is to make sure only the reset/prune tests are
using a custom network dir. This makes sure they do not wipe configs
that are otherwise required by other parallel running tests.
Fixes#17946
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
When creating storage for a container using ID maps, read the ID maps
that are assigned to the container from the returned container
structure, rather than from the options structure that we passed to the
storage library, which it previously modified in error.
[NO NEW TESTS NEEDED]
Signed-off-by: Nalin Dahyabhai <nalin@redhat.com>
Kube generate on pods was not checking for any underscores
in the pod name so was creating a kube yaml with an invalid
pod name when there were underscores present.
The hostname for the pod is set to the podname by default. There
is no need to set that to the container's name or the pod name
again in the generated yaml. So removed that field unless a hostname
was set for the container by the user.
Signed-off-by: Urvashi Mohnani <umohnani@redhat.com>
Fixes: https://github.com/containers/podman/issues/18040
If the `build_aarch64` task happens to fail for any reason, it will
cause the `curl` command in the `clone_script` for the aarch64 system
test tasks to throw a 404. This is because the
`local_system_test_aarch64_task` depends on `build` not `build_aarch64`.
As discovered in another issue long ago, the Cirrus API depends on doing
some dependency-resolution magic to function properly. Fix this by
correcting the dependencies.
Signed-off-by: Chris Evich <cevich@redhat.com>
Mention that specified credentials are only used to authenticate against
target registries (e.g., during `pull` or `build`) and are not used to
authenticat against mirrors etc.
Closes: #17185
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
Test has been flaking. Reason: container was run with -d, so
there's a small window in which podman-healthcheck ran on
a running container. Solution: remove -d
Signed-off-by: Ed Santiago <santiago@redhat.com>
Check the DELETE reports for both deletes. #18041 indicates that the
pod hasn't been removed which made me suspicious about the 1st delete.
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
The standard lib states that server handlers don't need to close the
body, so let's not do that to avoid any unforeseen side effect.
[NO TESTS NEEDED] - existing tests should suffice.
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
I recently wanted to check which version of Quadlet I was using and
found a `-version` flag to be missing. Since Quadlet and Podman are
bundled together, it seems reasonable to me for them to share the same
version.
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
Purely cosmetic change for consistency.
[NO NEW TESTS NEEDED] as it doesn't change functionality.
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
This commit adds an quadlet option `Tmpfs` which can be used to mount a
tmpfs in the container.
Closes#17907
Signed-off-by: Cedric Staniewski <cedric@gmx.ca>
I made a change in c/common[1] to prevent duplicates in netns names.
This now causes problem in podman[2] where the rootless netns will no
longer work after the netns got invalid but the underlying path still
exists. AFAICT this happens when the podman pause process got killed and
we are now in a different user namespace.
While I do not know what causes this, this commit should make it at
least possible to recover from this situation automatically as it used
to be before[1].
the problem with that is that containers started before it will not be
able to talk to contianers started after this. A restart of the previous
container will fix it but this was also the case before.
[NO NEW TESTS NEEDED]
[1] https://github.com/containers/common/pull/1381
[2] https://github.com/containers/podman/issues/17903#issuecomment-1494169843
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
One of our oldest most frustrating flakes is #16091, "Timed
out waiting for BYE".
In #17489 we added some debug output to see if the problem
was a container hang of some sort. It does not seem to be
(see #17675), and the debug output makes it hard to read
failure logs, so let's remove it.
Signed-off-by: Ed Santiago <santiago@redhat.com>
Add the command along with the abi and tunnel support
Add e2e tests
Add man page
Add apiv2 test to ensure return codes
Signed-off-by: Ygal Blum <ygal.blum@gmail.com>
MSI Validation will fail if the process env is > 32k chars
Remove CIRRUS_COMMIT_MESSAGE and CIRRUS_PR_BODY which can easily exceed this limit
[NO NEW TESTS NEEDED]
Signed-off-by: Jason T. Greene <jason.greene@redhat.com>
