The standard Docker client passes "1" for pull query parameter.
This means our compat endpoint has to work with this.
The endpoint was recently modified to accept pull-policy however this is
not how Docker actually work. Docker actually treats `pull` as boolean.
For sake of compatibility I decide to preserve pull-policy parsing too.
We can consider this podman's extension. I should not affect standard
clients.
Signed-off-by: Matej Vasek <mvasek@redhat.com>
We never ever close the stream so we do not need the Close() function in
th ebackend, the caller should close when required which may no be the
case, i.e. when os.Stdout/err is used.
This should not be a breaking change as the io.Writer is a subset of
io.WriteCloser, therfore all code should still compile while allowing to
pass in Writers without Close().
This is useful for podman top where we exec ps in the container via
podman exec.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
This ended up more complicated then expected. Lets start first with the
problem to show why I am doing this:
Currently we simply execute ps(1) in the container. This has some
drawbacks. First, obviously you need to have ps(1) in the container
image. That is no always the case especially in small images. Second,
even if you do it will often be only busybox's ps which supports far
less options.
Now we also have psgo which is used by default but that only supports a
small subset of ps(1) options. Implementing all options there is way to
much work.
Docker on the other hand executes ps(1) directly on the host and tries
to filter pids with `-q` an option which is not supported by busybox's
ps and conflicts with other ps(1) arguments. That means they fall back
to full ps(1) on the host and then filter based on the pid in the
output. This is kinda ugly and fails short because users can modify the
ps output and it may not even include the pid in the output which causes
an error.
So every solution has a different drawback, but what if we can combine
them somehow?! This commit tries exactly that.
We use ps(1) from the host and execute that in the container's pid
namespace.
There are some security concerns that must be addressed:
- mount the executable paths for ps and podman itself readonly to
prevent the container from overwriting it via /proc/self/exe.
- set NO_NEW_PRIVS, SET_DUMPABLE and PDEATHSIG
- close all non std fds to prevent leaking files in that the caller had
open
- unset all environment variables to not leak any into the contianer
Technically this could be a breaking change if somebody does not
have ps on the host and only in the container but I find that very
unlikely, we still have the exec in container fallback.
Because this can be insecure when the contianer has CAP_SYS_PTRACE we
still only use the podman exec version in that case.
This updates the docs accordingly, note that podman pod top never falls
back to executing ps in the container as this makes no sense with
multiple containers so I fixed the docs there as well.
Fixes#19001
Fixes https://bugzilla.redhat.com/show_bug.cgi?id=2215572
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
The libpod API does not set a default. Also PodTop is podman sepecific
so we can just rmeove this extra branch there.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Fixes a bug where `podman volume ls` with multiple `label` filters would
return volumes that matched *any* of the filters, not *all* of them.
Adapts generating volume filter functions to be more in
line with how it is done for containers and pods.
Fixes: #19219
Signed-off-by: Jake Correnti <jakecorrenti+github@proton.me>
When using 'podman run --rootfs ...', the image passed to SpecGenToOCI
may be nil - in this case, fall back to "freebsd" for the container OS.
[NO NEW TESTS NEEDED]
Signed-off-by: Doug Rabson <dfr@rabson.org>
During the exponential backoff waiting for the machine to be fully up
and running, also make sure that SSH is ready. The systemd dependencies
of the ready.service include the sshd.service among others but that is
not enough.
Other CoreOS users reported the same issue on IRC, so I feel fairly
confident to use the pragmatic approach of making sure SSH works on the
client side. #17403 is quite old and there are other pressing machine
issues that need attention.
[NO NEW TESTS NEEDED]
Fixes: #17403
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
Make sure that starting a qemu machine uses proper exponential backoffs
and that a single variable isn't shared across multiple backoffs.
DO NOT BACKPORT: I want to avoid backporting this PR to the upcoming 4.6
release as it increases the flakiness of machine start (see #17403). On
my M2 machine, the flake rate seems to have increased with this change
and I strongly suspect that additional/redundant sleep after waiting for
the machine to be running and listening reduced the flakiness. My hope
is to have more predictable behavior and find the sources of the flakes
soon.
[NO NEW TESTS NEEDED] - still too flaky to add a test to CI.
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
The --authfile flag has been ignored. Fix that and add a test to make
sure we won't regress another time. Requires a new --tls-verify flag
to actually test the code.
Also bump c/common since common/pull/1538 is required to correctly check
for updates. Note that I had to use the go-mod-edit-replace trick on
c/common as c/buildah would otherwise be moved back to 1.30.
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=2218315
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
When I reworked pod removal to provide more detailed errors
(including per-container errors, not just a single multierror
with all errors squashed), I made it part of the struct returned
by the REST API and assumed that would be enough to get errors
through to clients. Unfortunately, in case of an overarching
error removing the pod (as any error with any container would
cause), we don't send the response struct that would include the
container errors - we just send a standardized REST error. We
could work around this with custom, potentially backwards
incompatible error handling for the REST pod delete endpoint, or
we could just do what was done before, and package up all the
errors in a multierror to send to the other side. Of those
options, the multierror seems far simpler.
Fixes#19159
Signed-off-by: Matt Heon <mheon@redhat.com>
This endpoint queried the same package versions twice causing it to be
slower than info. Because it already called info we can just reuse the
package versions from there.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Previous tests have worked by pure chance since the client and server
ran on the same host; the server picked up the credentials created by
the client login.
Extend the gating tests and add a new integration test which is further
capable of exercising the remote code.
Note that fixing authentication support requires adding a new
`--authfile` CLi flag to `manifest inspect`. This will at least allow
for passing an authfile to be bindings. Username and password are not
yet supported.
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
Previously podman was using "MB" and "GB" (binary) for input but
"MB" and "GB" (decimal) for output, which was causing confusion.
Signed-off-by: Anders F Björklund <anders.f.bjorklund@gmail.com>
When working on Linux emulation on FreeBSD, I assumed that
SpecGenerator.ImageOS was always populated from the image's OS value but
in fact, this value comes from the CLI --os flag if set, otherwise "".
This broke running FreeBSD native containers unless --os=freebsd was
also set. Fix the problem by getting the value from the image itself.
This is a strong incentive for me to complete a stalled project to enable
podman system tests on FreeBSD.
[NO NEW TESTS NEEDED]
Signed-off-by: Doug Rabson <dfr@rabson.org>
This adds define.BindOptions to declare the mount options for bind-like
mounts (nullfs on FreeBSD). Note: this mirrors identical declarations in
buildah and it may be preferable to use buildah's copies throughout
podman.
[NO NEW TESTS NEEDED]
Signed-off-by: Doug Rabson <dfr@rabson.org>
There was a huge cut and paste of mount options which were not constent
in parsing tmpfs, bind and volume mounts. Consolidated into a single
function to guarantee all parse the same.
Fixes: https://github.com/containers/podman/issues/18995
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
This is limited to images that don't depend on complex cgroup or capability
setups but does cover enough functionality to be useful.
[NO NEW TESTS NEEDED]
Signed-off-by: Doug Rabson <dfr@rabson.org>
we were silently ignoring --device-cgroup-rule in rootless mode. Make
sure an error is returned if the user tries to use it.
Closes: https://github.com/containers/podman/issues/18698
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
Handle more TOCTOUs operating on listed images. Also pull in
containers/common/pull/1520 and containers/common/pull/1522 which do the
same on the internal layer tree.
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=2216700
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
Make sure we use the config field to know if we should use pasta or
slirp4netns as default.
While at it fix broken code which sets the default at two different
places, also do not set in Validate() as this should not modify the
specgen IMO, so set it directly before that.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
After[1] c/image no longer prints "Storing signatures" so we should
not check for it.
[1] https://github.com/containers/image/pull/2001
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
GetKeepIDMapping never read the gid (as it intended) but reused the uid.
Most likely a typo that never bothered anybody as uid and gid usually
match.
Signed-off-by: Simon Brakhane <simon@brakhane.net>
When debugging #17403, the logs of sshd indicates that Podman tried to
ssh into the machine too soon as the `core` user has not yet been fully
set up:
> error: kex_exchange_identification: Connection closed by remote host
> fatal: Access denied for user core by PAM account configuration [preauth]
@dustymabe found that the we may have to wait for systemd-user sessions
to be up. Doing that reduces the flake rate on my M2 machine but does
not entirely fix the issue.
Since I have seen multiple symptoms of flakiness, I think it does not
hurt to add the systemd-user sessions to the dependencies of the ready
service and continue investigating.
[NO NEW TESTS NEEDED] - once we have a fix out, I want to exercise
frequent stop/start in the machine tests but they won't pass now.
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
This adds the 'system service' command to the build on FreeBSD and
suppresses the call to servicereaper.Start which is only needed to
support slirp4netns on Linux. A stub for compat.StatsContainer is also
added - stats are still supported via the libpod.StatsContainer API
call.
[NO NEW TESTS NEEDED]
Signed-off-by: Doug Rabson <dfr@rabson.org>
Somehow my error message in top was never printed for the compat API,
the libpod one using the same code worked fine. Turns out the compat one
is using this buffered writter instaed but never made sure to flush it
before closing the connection.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Wait before sending status code 200 for the first top call and if that
fails return a proper error code.
This was leading to some confusion in [1] because podman just reported
200 but did not wirte anything back.
[1] https://bugzilla.redhat.com/show_bug.cgi?id=2215572
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Support two new wait conditions, "healthy" and "unhealthy". This
further paves the way for integrating sdnotify with health checks which
is currently being tracked in #6160.
Fixes: #13627
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
Massage the internal APIs to use a string slice instead of a state slice
for passing wait conditions. This paves the way for waiting on
non-state conditions such as "healthy".
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
Most of the code moved there so if from there and remove it here.
Some extra changes are required here. This is a bit of a mess. The pipe
handling makes this a bit more difficult.
[NO NEW TESTS NEEDED] This is just a rework, existing tests must pass.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Make sure that the create endpoint does not always return 200 even in
case of a failure. Some of the code had to be massaged since encoding a
report implies sending a 200.
Fixes: #15828
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
Make sure that the push endpoint does not always return 200 even in case
of a push failure. Some of the code had to be massaged since encoding a
report implies sending a 200.
Fixes: #18751
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
