podman wait by default waits for exit not removal as the man page
documents.
Fixes: 3a98b6dc0e ("test: Wait for killed container to avoid leak")
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Bash will expand a signle ? to a file name which consists of a single
char, and thus if you have a file named "a" in the cwd it will add a as
argument which causes podman a ... to be executed which clearly fails
the test.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
* Added flags to point to TLS PEM files to use for exposing and connecting
to an encrypted remote API socket with server and client authentication.
* Added TLS fields for system connection ls templates.
* Added special "tls" format for system connection ls to list TLS fields
in human-readable table format.
* Updated remote integration and system tests to allow specifying a
"transport" to run the full suite against a unix, tcp, tls, or mtls
system service.
* Added system tests to verify basic operation of unix, tcp, tls, and mtls
services, clients, and connections.
Signed-off-by: Andrew Melnick <meln5674.5674@gmail.com>
The distro-integration tag was added for fedora openQA to only run a
subset of tests. However since it was added only a few new tests have
been labelled like that and in general a normal contributor or even
maintianer has no idea when to add this tag.
We also have been seeing several regressions getting into fedora that
these tests would have caught. As such I worked with Adam to enable all
tests for fedora openQA so we actually have proper coverage. This has
been working for a few weeks so I think we can dop these tags so
upstream does not need to bother with them at all.
https://pagure.io/fedora-qa/os-autoinst-distri-fedora/issue/373
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Where possible, use safename and add ci:parallel tags.
One test runs "podman kill -a", which would be unwise to run
in parallel with other tests.
Signed-off-by: Ed Santiago <santiago@redhat.com>
The current logic used podman logs I don't understand way, all we care
about is the container output and we can just read the same with a
attached podman run, of course we have to move it into the background
but it did the some with logs.
This also allows us to remove the extra log-driver checks and because
podman logs seems to be much slower than the extra run we safe over 10s
with this change.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Also add a new `StoppedByUser` field to the container-inspect state
which can be useful during debugging and is now also used in the
regression test. Note that I moved the `false` check one test above
such that we can compare the previous Podman version which should just
be stuck in the `wait $ctr` command since it will continue restarting.
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
BATS 1.8.0 introduces tags: metadata that can be applied to
a single test or one entire file, then used for filtering
in a test run.
Issue #19299 introduces the possibility of using OpenQA
for podman reverse dependency testing: continuous CI on
all packages that can affect podman, so we don't go two
months with no bodhi builds then get caught by surprise
when systemd or kernel or crun change in ways that break us.
This PR introduces one bats tag, "distro-integration".
The intention is for OpenQA (or other) tests to install
the podman-tests package and run:
bats --filter-tags distro-integration /usr/share/podman/test/system
Goal is to keep the test list short and sweet: we do not
need to test command-line option parsing. We *DO* need to
test interactions with systemd, kernel, nethack, and other
critical components.
Signed-off-by: Ed Santiago <santiago@redhat.com>
Having a container spin-wait on a /stop file, then exit, is
unsafe: 'podman exec $ctr touch /stop' can get sucked into
container cleanup before the exec terminates, resulting in
the podman-exec failing and hence the test failing.
Most existing instances of this pattern are unnecessary.
Replace those with just 'podman rm -f'.
When necessary, use a variety of safer alternatives.
Re-Closes: #10825 (already closed; this addresses remaining cases)
Signed-off-by: Ed Santiago <santiago@redhat.com>
Make sure `Sync()` handles state transitions and exit codes correctly.
The function was only being called when batching which could render
containers in an unusable state when running concurrently with other
state-altering functions/commands since the state must be re-read from
the database before acting upon it.
Fixes: #14761
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
Make sure that `kill -a` is printing the IDs of killed containers.
Previously, it was only printing empty lines.
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
`logs -f` with `journald` is supported only when `journald` events
backend is used. To pass system tests using `logs -f` in an environment
where `events_logger` is not set to `journald` in `containers.conf`,
this fix sets `--events-backend` or `--log-driver` temporally.
Signed-off-by: Hironori Shiina <shiina.hironori@jp.fujitsu.com>
Increase the amount of time we're willing to wait for a log message that
a container should be printing to show up in the output of `logs -f`,
since on at least one CI configuration we're seeing a turnaround as high
as 46s, but it's not something we can directly control, so that's not a
hard maximum.
Signed-off-by: Nalin Dahyabhai <nalin@redhat.com>
Fix a race in journald driver. Following the logs implies streaming
until the container is dead. Streaming happened in one goroutine,
waiting for the container to exit/die and signaling that event happened
in another goroutine.
The nature of having two goroutines running simultaneously is pretty
much the core of the race condition. When the streaming goroutines
received the signal that the container has exitted, the routine may not
have read and written all of the container's logs.
Fix this race by reading both, the logs and the events, of the container
and stop streaming when the died/exited event has been read. The died
event is guaranteed to be after all logs in the journal which guarantees
not only consistencty but also a deterministic behavior.
Note that the journald log driver now requires the journald event
backend to be set.
Fixes: #10323
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
The initial version of libimage changed the order of layers which has
now been restored to remain backwards compatible.
Further changes:
* Fix a bug in the journald logging which requires to strip trailing
new lines from the message. The system tests did not pass due to
empty new lines. Triggered by changing the default logger to
journald in containers/common.
* Fix another bug in the journald logging which embedded the container
ID inside the message rather than the specifid field. That surfaced
in a preceeding whitespace of each log line which broke the system
tests.
* Alter the system tests to make sure that the k8s-file and the
journald logging drivers are executed.
* A number of e2e tests have been changed to force the k8s-file driver
to make them pass when running inside a root container.
* Increase the timeout in a kill test which seems to take longer now.
Reasons are unknown. Tests passed earlier and no signal-related
changes happend. It may be CI VM flake since some system tests but
other flaked.
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
- images test: add test for 'table' and '\t' formatting
- image mount test: check output from 'umount', test
repeat umount (NOP), and test invalid-umount
- kill test: remove kludgy workaround for crun signal bug
ref: #5004 -- code is no longer needed (fingers crossed),
and the workaround involved pulling an expensive image.
- selinux test: add new tests for shared context in:
* pods , w/ and w/o infra container (ref: #7902)
* containers with namespace sharing: --ipc, --pid, --net
- selinux test: new test for --pid=host (disabled pending
propagation of container-selinux-2.146, ref: #7939)
Signed-off-by: Ed Santiago <santiago@redhat.com>
Invert the branch logic to match the comment. Docker seems to wait for
the container while Podman does not.
Enable the remote-disabled system test as well.
Fixes: #7135
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
podman-remote is in better shape now. Let's see what needs
to be done to reenable remote system tests.
- logs test: skip multilog, it doesn't work remote
- diff test: use -l only when local, not with remote
- many other tests: skip_if_remote, with 'FIXME: pending #xxxx'
where xxxx is a filed issue.
Unrelated: added new helper to skip_if_remote and _if_rootless,
where we check if the source message includes "remote"/"rootless"
and insert it if missing. This is a minor usability enhancement
to make it easier to understand at-a-glance why a skip triggers.
Signed-off-by: Ed Santiago <santiago@redhat.com>
9f69c4eca (part of the f31 pr, #3091) semi-broke the kill test,
there's now an ugly warning:
setup(): removing stray images quay.io/libpod/fedora-minimal:latest 7bb5a60e8a78
The comments also didn't actually explain the problem
being addressed, and included a misleading reference
to busybox.
Here we switch to using fedora-minimal only with podman-remote,
clean it up (rmi) when finished, and include an explanation in
the comments about why this is needed; making it clear that
this workaround can be removed once we get rid of podman-remote.
We also reformat back to 80 columns.
Signed-off-by: Ed Santiago <santiago@redhat.com>
It's possible/likely the container image for the test will need to be
pulled as part of the `run` command. Due to the way BATS handles
output, messages regarding image-pull could be misinterpreted as the
container's CID. Force the CID to be obtained by only the last line of
output.
Signed-off-by: Chris Evich <cevich@redhat.com>
Add pkg/signal to deal with parts of signal processing and translating
signals from string to numeric representations. The code has been
copied from docker/docker (and attributed with the copyright) but been
reduced to only what libpod needs (on Linux).
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
When you open a FIFO for reading, but there's no writer, you hang.
This is just one of those obscure UNIXisms we all know but just
forget all too often.
My last PR was guilty of introducing such a condition; I caught
it by accident while testing other stuff. In short, the signal
container was doing 'echo DONE' as its last step, and we (BATS)
were reading the FIFO to check for it; but if the container
exited before we opened the FIFO for read, the open would hang.
This is not a hang that we can catch in the test: it would hang
the entire job forever. CI would presumably time out eventually,
but with no useful indication of the cause of the error.
Solution: use 'exec' to open the FIFO early and keep it open,
and use 'read -u FD' instead of 'read <$fifo': the former
reads from an open FD, the latter forces a new open() each time.
There is a shorter, more maintainable solution -- see #4755 -- but
that suffers from the same hanging problem in the (unlikely) case
where the signal-handling container exits, e.g. if signal handling
is broken in podman. The test would hang, with no helpful indicator.
Although this PR is a little more advanced scripting, I have
commented the relevant code well and believe the maintenance
cost is worth the risk of undebuggable hangs.
There is still a hang risk: if 'podman logs -f' fails and exits
immediately, the 'exec' will hang. I can't think of a non-racy
way to prevent that, and choose to live with that risk.
Tested by temporarily including 9 (SIGKILL) in the signals list.
The read timeout triggers, and the end user has a fair chance
of tracking down the root cause.
Signed-off-by: Ed Santiago <santiago@redhat.com>
The helper function we use for signal name mapping does not
check for negative numbers nor invalid (too-high) ones. This
can yield unexpected error messages:
# podman kill -s -1 foo
ERRO[0000] unknown signal "18446744073709551615"
This PR introduces a small wrapper for it that:
1) Strips off a leading dash, allowing '-1' or '-HUP'
as valid inputs; and
2) Rejects numbers <1 or >64 (SIGRTMAX)
Also adds a test suite checking signal handling as well as
ensuring that invalid signals are rejected by the command line.
Fixes: #4746
Signed-off-by: Ed Santiago <santiago@redhat.com>
