When Libpod removes a container, there is the possibility that
removal will not fully succeed. The most notable problems are
storage issues, where the container cannot be removed from
c/storage.
When this occurs, we were faced with a choice. We can keep the
container in the state, appearing in `podman ps` and available for
other API operations, but likely unable to do any of them as it's
been partially removed. Or we can remove it very early and clean
up after it's already gone. We have, until now, used the second
approach.
The problem that arises is intermittent problems removing
storage. We end up removing a container, failing to remove its
storage, and ending up with a container permanently stuck in
c/storage that we can't remove with the normal Podman CLI, can't
use the name of, and generally can't interact with. A notable
cause is when Podman is hit by a SIGKILL midway through removal,
which can consistently cause `podman rm` to fail to remove
storage.
We now add a new state for containers that are in the process of
being removed, ContainerStateRemoving. We set this at the
beginning of the removal process. It notifies Podman that the
container cannot be used anymore, but preserves it in the DB
until it is fully removed. This will allow Remove to be run on
these containers again, which should successfully remove storage
if it fails.
Fixes#3906
Signed-off-by: Matthew Heon <mheon@redhat.com>
Add a --cidfile flag to podman rm/stop to pass a container ID via a
file. Podman run already provides the functionaly to store the ID
in a specified file which we now complete with rm/stop. This allows
for a better life-cycle management in systemd services. Note that
--cdifile can be specified multiple times to rm/stop.
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
When restoring a container with user namespace, the user namespace is
created by the OCI runtime, and the network namespace is created after
the user namespace to ensure correct ownership.
In this case PostConfigureNetNS will be set and the value of
c.state.NetNS would be nil. Hence, the following error occurs:
$ sudo podman run --name cr \
--uidmap 0:1000:500 \
-d docker.io/library/alpine \
/bin/sh -c 'i=0; while true; do echo $i; i=$(expr $i + 1); sleep 1; done'
$ sudo podman container checkpoint cr
$ sudo podman container restore cr
...
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x30 pc=0x13a5e3c]
Signed-off-by: Radostin Stoyanov <rstoyanov1@gmail.com>
Most build testing should be done in Buildah's test
suites, but we should have a minimal amount of tests,
especially testing the parts that are different like
layers and squash. Also the CLI argument handling
of things like the context directory that we've had
issues reported.
This first chunk does a basic test and then checks for
context directory being a file and squash iterations.
More to be added as time goes by.
Signed-off-by: TomSweeneyRedHat <tsweeney@redhat.com>
In conmon 2.0.3, we add another fifo to handle window resizing. This needs to be cleaned up for commands like restore, where the same path is used.
Signed-off-by: Peter Hunt <pehunt@redhat.com>
the pull all tags test can frequently timeout when trying to pull all
alpine tags. using the pause image, which is smaller, should provide
some relief.
Signed-off-by: baude <bbaude@redhat.com>
timetype package include functions for calculating
time input in string as one of golang duration format and
calculate as unix timestamp
Signed-off-by: Kunal Kushwaha <kunal.kushwaha@gmail.com>
Bump to Buildah v1.11.5. Most notably changes to the
podman build `--pull` functionality. `--pull=true` and `--pull=false` now
work as Docker does, `--pull-never` added to supply the functionality
of the old `--pull=false`.
Signed-off-by: TomSweeneyRedHat <tsweeney@redhat.com>
Unless explicitely overridden, check if the image's OS and architecture
and throw an errors in case of a mismatch.
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
Rewrite the backend for displaying the history of an image to simplify
the code and be closer to docker's behaviour. Instead of driving
index-based heuristics, create a reverse mapping from top-layers to the
corresponding image IDs and lookup the layers on-demand. Also use the
uncompressed layer size to be closer to Docker's behaviour.
Note that intermediate images from local builds are not considered for
the ID lookups anymore.
Fixes: #3359
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
Automated building of container images is handled in quay.io based on
changes in the master branch of this repository. However, as additional
branches are made, the "latest" image (from master) diverges from their
expectations. Fix this by using the branch-tagged images built by quay.
For the near-term, this also implies quay.io will be configured to also
build different images for each branch, and tag them appropriately.
Long-term, image build automation should be combined with libpod
automation - to avoid needing to maintain automation in multiple
systems/locations.
Signed-off-by: Chris Evich <cevich@redhat.com>
Warning message added before executing image prune
Added a force option, to execute without user input.
Signed-off-by: Kunal Kushwaha <kunal.kushwaha@gmail.com>
podman in Fedora gets seccomp.json from containers-common while
the one in Ubuntu PPA gets seccomp.json from containers-golang.
This change will let me use install.config target unmodified
in downstream packages.
Signed-off-by: Lokesh Mandvekar <lsm5@fedoraproject.org>
We have the annotations SandboxID, let's use them. This also allows kata containers to be created in pods and share a VM with the infra container. Note: as of now, this sharing only works if the pod has an infra container
Signed-off-by: Peter Hunt <pehunt@redhat.com>
Add missing options.
Define the use of the TMPDIR environment variable for the storing of container images
when they are being pulled to the system.
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
As it stands, createconfig is a huge struct. This works fine when the only caller is when we create a container with a fully created config. However, if we wish to share code for security and namespace configuration, a single large struct becomes unweildy, as well as difficult to configure with the single createConfigToOCISpec function.
This PR breaks up namespace and security configuration into their own structs, with the eventual goal of allowing the namespace/security fields to be configured by the pod create cli, and allow the infra container to share this with the pod's containers.
Signed-off-by: Peter Hunt <pehunt@redhat.com>
