Because the test left the image mounted the cleanup failed to remove the
tmpdir as it contained an active mount point. Thus ensure we unmount the
image again to prevent this leak.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Using /tmp means this file will be leaked and no deleted, switch to
using the per test tempdir which is removed after the test.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
TMPDIR is typically /tmp which is typically(*) a tmpfs.
This PR ignores $TMPDIR when $CI is defined, forcing all
e2e tests to set up one central working directory in /var/tmp
instead.
Also, lots of cleanup.
(*) For many years, up to and still including the time of
this PR, /tmp on Fedora CI VMs is actually NOT tmpfs,
it is just / (root). This is nonstandard and undesirable.
Efforts are underway to remove this special case.
Signed-off-by: Ed Santiago <santiago@redhat.com>
It is not clear why rootless was forced to the cgroupfs manager when
systemd is the default. In any case it causes local test failures as
described in the issue[1]. Using systemd manager makes them pass as
expected, I don't know enough aout cgroups to know the difference and
why certain tests have bad asumptions but this fixes it.
[1] https://github.com/containers/podman/issues/22474
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
When the source dir is already mounted noexec, nodev or nosuid then a
rootless user cannot mount the dir into the container without these
options for obvious reasons.
So in order to run the test we must ensure the dir is mounted with these
options first, if they are simply skip as the test will fail otherwise.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
When you run locally with a higher oom_score_adj then the one used in
the test podman will print a warning and not set the oom lower then the
current value. Thus use 999 as value which should only cause problems
for users with oom_score_adj value of 1000 (max value) which seems
unlikely.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Effectively, this is an ability to take an image already pulled
to the system, and automatically mount it into one or more
containers defined in Kubernetes YAML accepted by `podman play`.
Requirements:
- The image must already exist in storage.
- The image must have at least 1 volume directive.
- The path given by the volume directive will be mounted from the
image into the container. For example, an image with a volume
at `/test/test_dir` will have `/test/test_dir` in the image
mounted to `/test/test_dir` in the container.
- Multiple images can be specified. If multiple images have a
volume at a specific path, the last image specified trumps.
- The images are always mounted read-only.
- Images to mount are defined in the annotation
"io.podman.annotations.kube.image.automount/$ctrname" as a
semicolon-separated list. They are mounted into a single
container in the pod, not the whole pod.
As we're using a nonstandard annotation, this is Podman only, any
Kubernetes install will just ignore this.
Underneath, this compiles down to an image volume
(`podman run --mount type=image,...`) with subpaths to specify
what bits we want to mount into the container.
Signed-off-by: Matt Heon <mheon@redhat.com>
Image volumes (the `--mount type=image,...` kind, not the
`podman volume create --driver image ...` kind - it's strange
that we have two) are needed for our automount scheme, but the
request is that we mount only specific subpaths from the image
into the container. To do that, we need image volume subpath
support. Not that difficult code-wise, mostly just plumbing.
Also, add support to the CLI; not strictly necessary, but it
doesn't hurt anything and will make testing easier.
Signed-off-by: Matt Heon <mheon@redhat.com>
Followup to [1]#22270: wherever possible/practical, extend command
error checks to include explicit exit status codes and error strings.
Just trying to shrink down #22346 to a manageable, reviewable size.
Signed-off-by: Ed Santiago <santiago@redhat.com>
Followup to #22270: wherever possible/practical, extend command
error checks to include explicit exit status codes and error strings.
More low-hanging fruit: small reviewable chunks
Signed-off-by: Ed Santiago <santiago@redhat.com>
Followup to #22270: wherever possible/practical, extend command
error checks to include explicit exit status codes and error strings.
Because #22346 is stalled, these are some trivial easy-to-review
changes that get us closer to the goal.
Signed-off-by: Ed Santiago <santiago@redhat.com>
When the field is set to false we should never log healthcheck events.
Fixes https://issues.redhat.com/browse/RHEL-18987
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
The change in healthcheck_run_test.go, depends on the
containers/image change:
commit b6afa8ca7b324aca8fd5a7b5b206fc05c0c04874
Author: Mikhail Sokolov <msokolov@evolution.com>
Date: Fri Mar 15 13:37:44 2024 +0200
Add support for Docker HealthConfig.StartInterval (v25.0.0+)
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
This is something Docker does, and we did not do until now. Most
difficult/annoying part was the REST API, where I did not really
want to modify the struct being sent, so I made the new restart
policy parameters query parameters instead.
Testing was also a bit annoying, because testing restart policy
always is.
Signed-off-by: Matt Heon <mheon@redhat.com>
The Docker endpoint here is kind of a nightmare - accepts a full
Resources block, including a large number of scary things like
devices. But it only documents (and seems to use) a small subset
of those. This implements support for that subset. We can always
extend things to implement more later if we have a need.
Signed-off-by: Matt Heon <mheon@redhat.com>
The logic here is more complex than I would like, largely due to
the behavior of `podman inspect` for running containers. When a
container is running, `podman inspect` will source as much as
possible from the OCI spec used to run that container, to grab
up-to-date information on things like devices. We don't want to
change this, it's definitely the right behavior, but it does make
updating a running container inconvenient: we have to rewrite the
OCI spec as part of the update to make sure that `podman inspect`
will read the correct resource limits.
Also, make update emit events. Docker does it, we should as well.
Signed-off-by: Matt Heon <mheon@redhat.com>
...and an optional error-message string, to be checked
against stderr.
This is a starting point and baby-steps progress toward #18188.
There are 249 ExitWithError() checks in test/e2e. It will take
weeks to fix them all. This commit enables new functionality:
Expect(ExitWithError(125, "expected substring"))
...while also allowing the current empty-args form. Once
all 249 empty-args uses are modernized, the matcher code
will be cleaned up.
I expect it will take several months of light effort to get
all e2e tests transitioned to the new form. I am choosing to
do so in pieces, for (relative) ease of review. This PR:
1) makes the initial changes described above; and
2) updates a small subset of e2e _test.go files such that:
a) ExitWithError() is given an exit code and error string; and
b) Exit(Nonzero) is changed to ExitWithError(Nonzero, "string")
(when possible)
Signed-off-by: Ed Santiago <santiago@redhat.com>
Docker shows exposed ports as just PORT/PROTO so match that behavior. It
is not clear to me why someone needs that information in ps as "expose"
doesn't effect anything networking related.
Fixes https://issues.redhat.com/browse/RHEL-32154
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
As of April 2024, it's no longer included in rawhide by default.
We could force-install it, but it's 2024 and it seems likely
that all systems on which Podman 5 will run will have kernels
that support native overlay.
I also added two debugging printfs to the 'podman info' test
that initially failed on an (unpublished) rawhide VM. Without
these printfs it was impossible to diagnose the failure.
Updating docs is left as a future exercise.
Signed-off-by: Ed Santiago <santiago@redhat.com>
if the 'U' option is provided, do not chown the destination target to
the existing target in the image.
Closes: https://github.com/containers/podman/issues/22224
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
there are no overlay mounts in the "podman run with --volume and U
flag" tests so no need to skip them.
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
Emergency update to get pasta 03-26. Also gives us crun 1.14.4.
One unexplained difference: fc39 and rawhide now create:
/run/log/journal/SOMETHING/system.journal
...and the SOMETHING is o-rwx. This triggers journalctl to spit out a warning:
Hint: You are currently not seeing messages from the system.
Users in groups 'adm', 'systemd-journal', 'wheel' can see all messages.
Pass -q to turn off this notice.
...which in turn causes ExitCleanly() to fail.
It is not clear who/what is creating this journal directory, or
why it allofasudden started just now. Workaround is to add -q
to journalctl in one test.
One more difference, another test now requires SYSLOG capability.
VM package info:
https://github.com/containers/automation_images/pull/342
Signed-off-by: Ed Santiago <santiago@redhat.com>
The annotations should be maintained by CRI-O itself to decouple the
projects from a dependency perspective.
[NO NEW TESTS NEEDED]
Signed-off-by: Sascha Grunert <sgrunert@redhat.com>
Most of them look like our usual "assume too much about run -d".
One of them is just an unexpected warning, a push retry. Remove
the ExitCleanly() from that test, just rely on Exit(0).
The other two have to do with podman logs, which we know can lag.
Add a short 1-second retry loop.
Signed-off-by: Ed Santiago <santiago@redhat.com>
Two test flakes in the past week. Looks like the usual race
between "run -d" and "assume the container is ready". I don't
know if this will resolve them, but it's still a good idea.
Signed-off-by: Ed Santiago <santiago@redhat.com>
There are many code paths which only do logrus but still exit 0 so this
should catch more bugs. Unfortunately runc logs way to much random stuff
so we ignore this check for runc right now.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
when performing a system reset with containers that run somewhere where
a soft kill wont work (like sleep), containers will wait 10 seconds
before terminating with a sigkill. But for a forceful action like
system reset, we should outright set no timeout so containers stop
quickly and are not waiting on a timeout
Fixes#21874
Signed-off-by: Brent Baude <bbaude@redhat.com>
This vendors the latest c/common version, including making Pasta
the default rootless network provider. That broke a number of
tests, which have been fixed as part of this PR.
Also includes a change to network stats logic, which simplifies
the code a bit and makes it actually work with Pasta.
Signed-off-by: Matt Heon <mheon@redhat.com>
fix an assumption in the test, that the --compression-format is
unchanged from the original image.
Instead validate that all the required architectures are part of the
manifest.
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
Like docker podman network inspect should output the information of
running container with their ip/mac address on this network.
However the output format is not docker compatible as this cannot
include all the info we have and the previous output was already not
compatible so this is not new.
New example output:
```
[
{
...
"containers": {
"7c0d295779cee4a6db7adc07a99e635909413a390eeab9f951edbc4aac406bf1": {
"name": "c2",
"interfaces": {
"eth0": {
"subnets": [
{
"ipnet": "10.89.0.4/24",
"gateway": "10.89.0.1"
},
{
"ipnet": "fda3:b4da:da1e:7e9d::4/64",
"gateway": "fda3:b4da:da1e:7e9d::1"
}
],
"mac_address": "1a:bd:ca:ea:4b:3a"
}
}
},
"b17c6651ae6d9cc7d5825968e01d6b1e67f44460bb0c140bcc32bd9d436ac11d": {
"name": "c1",
"interfaces": {
"eth0": {
"subnets": [
{
"ipnet": "10.89.0.3/24",
"gateway": "10.89.0.1"
},
{
"ipnet": "fda3:b4da:da1e:7e9d::3/64",
"gateway": "fda3:b4da:da1e:7e9d::1"
}
],
"mac_address": "f6:50:e6:22:d9:55"
}
}
}
}
}
]
```
Fixes#14126
Fixes https://issues.redhat.com/browse/RHEL-3153
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Too many tests use port 5000. Although there's a putative GetPortLock()
it seems to be unreliable, and we often get what appear to be collisions
between tests.
A proper solution would be to pseudorandomly allocate ports, verify
that they're not being reused, Sprintf() these everywhere that
needs them, and sprinkle some powdered cinnamon on top.
This is not that proper solution.
Fixes: #20655
Signed-off-by: Ed Santiago <santiago@redhat.com>
