When doing a podman images, manifests lists look just like images, so
it is logical that users would assume that they can just podman push them
to a registry. The problem is we throw out weird errors when this happens
and users need to somehow figure out this is a manifest list rather then
an image, and frankly the user will not understand the difference.
This PR will make podman push just do the right thing, by failing over and
attempting to push the manifest if it fails to push the image.
Fix up handling of manifest push
Protocol should bring back a digest string, which can either be
printed or stored in a file.
We should not reimplement the manifest push setup code in the tunnel
code but take advantage of the api path, to make sure remote and local
work the same way.
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
partially revert 95c45773d7
restrict the cases where /sys is bind mounted from the host.
The heuristic doesn't detect all the cases where the bind mount is not
necessary, but it is an improvement on the previous version where /sys
was always bind mounted for rootless containers unless --net none was
specified.
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
Basic theory: We remove the container, but *only from the DB*.
We leave it in c/storage, we leave the lock allocated, we leave
it running (if it is). Then we create an identical container with
an altered name, and add that back to the database. Theoretically
we now have a renamed container.
The advantage of this approach is that it doesn't just apply to
rename - we can use this to make *any* configuration change to a
container that does not alter its container ID.
Potential problems are numerous. This process is *THOROUGHLY*
non-atomic at present - if you `kill -9` Podman mid-rename things
will be in a bad place, for example. Also, we can't rename
containers that can't be removed normally - IE, containers with
dependencies (pod infra containers, for example).
The largest potential improvement will be to move the majority of
the work into the DB, with a `RecreateContainer()` method - that
will add atomicity, and let us remove the container without
worrying about depencies and similar issues.
Potential problems: long-running processes that edit the DB and
may have an older version of the configuration around. Most
notable example is `podman run --rm` - the removal command needed
to be manually edited to avoid this one. This begins to get at
the heart of me not wanting to do this in the first place...
This provides CLI and API implementations for frontend, but no
tunnel implementation. It will be added in a future release (just
held back for time now - we need this in 3.0 and are running low
on time).
This is honestly kind of horrifying, but I think it will work.
Signed-off-by: Matthew Heon <mheon@redhat.com>
This involves a new test binary (a basic implementation of the
volume plugin protocol) and a new image on quay.io (Containerfile
to produce it and all sources located in this commit). The image
is used to run a containerized plugin we can test against.
Signed-off-by: Matthew Heon <mheon@redhat.com>
This implements support for mounting and unmounting volumes
backed by volume plugins. Support for actually retrieving
plugins requires a pull request to land in containers.conf and
then that to be vendored, and as such is not yet ready. Given
this, this code is only compile tested. However, the code for
everything past retrieving the plugin has been written - there is
support for creating, removing, mounting, and unmounting volumes,
which should allow full functionality once the c/common PR is
merged.
A major change is the signature of the MountPoint function for
volumes, which now, by necessity, returns an error. Named volumes
managed by a plugin do not have a mountpoint we control; instead,
it is managed entirely by the plugin. As such, we need to cache
the path in the DB, and calls to retrieve it now need to access
the DB (and may fail as such).
Notably absent is support for SELinux relabelling and chowning
these volumes. Given that we don't manage the mountpoint for
these volumes, I am extremely reluctant to try and modify it - we
could easily break the plugin trying to chown or relabel it.
Also, we had no less than *5* separate implementations of
inspecting a volume floating around in pkg/infra/abi and
pkg/api/handlers/libpod. And none of them used volume.Inspect(),
the only correct way of inspecting volumes. Remove them all and
consolidate to using the correct way. Compat API is likely still
doing things the wrong way, but that is an issue for another day.
Fixes#4304
Signed-off-by: Matthew Heon <matthew.heon@pm.me>
We now set Entrypoint when interpeting the image Entrypoint (or yaml.Command)
and Command when interpreting image Cmd (or yaml.Args)
This change is kind of breaking because now checking Config.Cmd won't return
the full command, but only the {cmd,args}.
Adapt the tests to this change as well
Signed-off-by: Peter Hunt <pehunt@redhat.com>
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
Adding another check in the `podman search --list-tags --format json` test case.
Replacing an anonymous struct by \`listEntryTag\` struct.
Signed-off-by: Alexandre Fourcat <afourcat@gmail.com>
`podman ps --format {{.Networks}}` will show all connected networks for
this container. For `pod ps` it will show the infra container networks.
Signed-off-by: Paul Holzinger <paul.holzinger@web.de>
Allow to filter on the network name or full id.
For pod ps it will filter on the infra container networks.
Signed-off-by: Paul Holzinger <paul.holzinger@web.de>
When I launch a container with --userns=keep-id the rootless processes
should have no caps by default even if I launch the container with
--privileged. It should only get the caps if I specify by hand the
caps I want leaked to the process.
Currently we turn off capeff and capamb, but not capinh. This patch
treats capinh the same way as capeff and capamb.
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
First, use the pflag library to parse the flags. With this we can
handle all corner cases such as -td or --detach=false.
Second, preserve the root args with --new. They are used for all podman
commands in the unit file. (e.g. podman --root /tmp run alpine)
Signed-off-by: Paul Holzinger <paul.holzinger@web.de>
When adding the HOSTNAME environment variable, only do so if it
is not already present in the spec. If it is already present, it
was likely added by the user, and we should honor their requested
value.
Fixes#8886
Signed-off-by: Matthew Heon <mheon@redhat.com>
`KillMode=none` has been deprecated in systemd and is now throwing big
warnings when being used. Users have reported the issues upstream
(see #8615) and on the mailing list.
This deprecation was mainly motivated by an abusive use of third-party
vendors causing all kinds of undesired side-effects. For instance, busy
mounts that delay reboot.
After talking to the systemd team, we came up with the following plan:
**Short term**: we can use TimeoutStopSec and remove KillMode=none which
will default to cgroup.
**Long term**: we want to change the type to sdnotify. The plumbing for
Podman is done but we need it for conmon. Once sdnotify is working, we
can get rid of the pidfile handling etc. and let Podman handle it.
Michal Seklatar came up with a nice idea that Podman increase the time
out on demand. That's a much cleaner way than hard-coding the time out
in the unit as suggest in the short-term solution.
This change is executing the short-term plan and sets a minimum timeout
of 60 seconds. User-specified timeouts are added to that.
Fixes: #8615
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
so that the PIDFile can be accessed also without being in the rootless
user namespace.
Closes: https://github.com/containers/podman/issues/8506
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
1.) podman cannot send proper dockerfile when it is not inside root
folder.
2.) support for sending symlinks and folders inside context dir
3.) when sending context dir as tar to remote, prevent sending items
inside .dockerignore
Signed-off-by: Kazım SARIKAYA <kazimsarikaya@sanaldiyar.com>
When running a privileged container and `SeccompProfilePath` is empty no seccomp profile should be applied.
(Previously this was the case only if `SeccompProfilePath` was set to a non-empty default path.)
Closes#8849
Signed-off-by: Max Goltzsche <max.goltzsche@gmail.com>
when HostNetwork is true in the pod spec.
Also propagate whether host network namespace should be used for containers.
Add test for HostNetwork setting in kubeYaml.
The infra configuration should reflect the setting.
Signed-off-by: Benedikt Ziemons <ben@rs485.network>
