Someone please revert this once quay search is fixed.
Signed-off-by: Ed Santiago <santiago@redhat.com>
<MH: Cherry-pick into v4.2.0-rhel to fix CI>
Signed-off-by: Matt Heon <mheon@redhat.com>
Docker's newer clients popuates `cacheFrom` and `cacheTo` parameter
by default as empty array for all commands but buildah's design of
distributed cache expects this to be a repo not image hence parse
only the first populated repo and igore if empty array.
Signed-off-by: Aditya R <arajan@redhat.com>
We are unmarshaling and re-marshaling JSON, which can _silently_ drop data
with the Go design decision.data.
Try harder, by using json.RawMessage at least for the data we care about.
Alternatively, this could use json.Decoder.DisallowUnknownFields.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
sigstoreSigned does not have GPG IDs, so we add N/A in that column.
NOTE: this does not show the use-sigstore-attachments value from
registries.d.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
Currently
- the output uses the first entry's type, even if the requirements are different
(notably signedBy + sigstoreSIgned)
- all public keys IDs are collected to a single line, even if some of them
are interchangeable, and some are required (e.g. two signedBy requirements
could require an image to be signed by (redhatProd OR redhatBeta) AND (vendor1 OR vendor2)
So, stop collapsing the requirements, and return a separate entry for each one. Multiple
GPG IDs on a single line used to mean AND or OR, now they always mean AND.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
Do the registries.d lookup once, separately from building
an entry, so that we can share it across entries.
Also prepare a separate res to allow adding multiple entries.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
Just so that we don't have a boolean-named function returning a struct.
Also reorder the parameters to have the container first, and the lookup
key second.
Shoud not change behavior.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
Now that it is the primary return value of a small function,
the long name only makes reading harder.
Should not change behavior.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
This will evetually allow us to use it for the default scope
as well, which currently uses a simplified version.
Should not change behavior.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
Add at least a basic unit test for the various entry types.
So that we don't have to actually deal with GPG keys and /usr/bin/gpg*,
parametrize the code with a gpgIDReader , and pass a fake one
in the unit test.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
We now have only a few entrypoints that are called externally,
so make the rest private. This will make it more obvious that
we are not breaking any external users.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
This will allow us to write unit tests without setting up the complete Podman runtime
(and without the Linux dependency).
Should not change behavior.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
NOTE: This does not edit the use-sigstore-attachments value
in registries.d, similarly to how (podman image trust set) didn't
set the lookaside paths for simple signing.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
- Also reject public keys with types that don't use them
- Reject unknown trust types
- And add unit tests
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
This will allow us to write unit tests without setting up the complete Podman runtime
(and without the Linux dependency).
Also, actually add a basic smoke test of the core functionality.
Should not change behavior.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
Only process the incoming args[] (which is a single-element array
for some reason) once, and use a semantic variable name for the value
we care about.
Should not change behavior, the only caller already supposedly ensures
that len(args) == 1.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
Split the existing code into policy.go and registries.go,
depending on which files it concerns.
Only moves unchanged code, should not change behavior.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
We can always recover it from git, but it seems to serve
no purpose anyway.
Should not change behavior.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
Also, do a general cleanup of all the timeout code. Changes
include:
- Convert from int to *uint where possible. Timeouts cannot be
negative, hence the uint change; and a timeout of 0 is valid,
so we need a new way to detect that the user set a timeout
(hence, pointer).
- Change name in the database to avoid conflicts between new data
type and old one. This will cause timeouts set with 4.2.0 to be
lost, but considering nobody is using the feature at present
(and the lack of validation means we could have invalid,
negative timeouts in the DB) this feels safe.
- Ensure volume plugin timeouts can only be used with volumes
created using a plugin. Timeouts on the local driver are
nonsensical.
- Remove the existing test, as it did not use a volume plugin.
Write a new test that does.
The actual plumbing of the containers.conf timeout in is one line
in volume_api.go; the remainder are the above-described cleanups.
Backported to v4.2.0-rhel per RHBZ 2125241
Signed-off-by: Matthew Heon <mheon@redhat.com>
Trying to print the image id on a failed inspect will result in a nil
pointer panic because the image will be nil. Replace image.id with the
image name which is defined as a string without the use of inspect.
Fixes: bz#2131836
[NO NEW TESTS NEEDED]
Signed-off-by: Brent Baude <bbaude@redhat.com>
For systems that have extreme robustness requirements (edge devices,
particularly those in difficult to access environments), it is important
that applications continue running in all circumstances. When the
application fails, Podman must restart it automatically to provide this
robustness. Otherwise, these devices may require customer IT to
physically gain access to restart, which can be prohibitively difficult.
Add a new `--on-failure` flag that supports four actions:
- **none**: Take no action.
- **kill**: Kill the container.
- **restart**: Restart the container. Do not combine the `restart`
action with the `--restart` flag. When running inside of
a systemd unit, consider using the `kill` or `stop`
action instead to make use of systemd's restart policy.
- **stop**: Stop the container.
To remain backwards compatible, **none** is the default action.
Backport of commit aad29e759c78
BZ: https://bugzilla.redhat.com/show_bug.cgi?id=2097708
Signed-off-by: Valentin Rothberg <vrothberg@redhat.com>
when verifying that the memory was set correctly for a podman machine
instance, we check if the number is between a range because based on
architecture, operating system, and memory itself this number can differ
significantly.
Signed-off-by: Brent Baude <bbaude@redhat.com>
use the sandbox id instead of the name for the
io.kubernetes.cri-o.SandboxID annotation used by gVisor.
Closes: https://github.com/containers/podman/issues/15223
[NO NEW TESTS NEEDED] it is specific to gVisor
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
Tying filtering logic for podman stop and start to same place in getContainersAndInputByContext() to reduce code redundancy
Signed-off-by: Karthik Elango <kelango@redhat.com>
podman run/create can accept `-h <hostname>` as argument. When parsing
flags -h throws an help requested error from pflag. To prevent this
error we have to define the help flag.
Fixes#15124
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
When a container was created with `--sdnotify value` we would remove
this arg instead of using it like with `--sdnotfiy=value`.
Also when the arg is set to ignore we should force conmon in order to
make the resulting Type=notify units work.
Fixes#15052
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
This bug is reproduced when we execute the following command:
1. podman manifest add <manifest list> <images exist on local storage>
2. podman manifest push --rm <manifest list> dir:<directory>
If pushing succeeds, it is expected to remove only a manifest list.
However, manifest list remains on local storage and images are removed.
This commit fixes `podman manifest push --rm` to remove only a manifest list.
And, supports `manifest push --rm option` in remote environment,
like host environment.
Fixes: https://github.com/containers/podman/issues/15033
Signed-off-by: Toshiki Sonoda <sonoda.toshiki@fujitsu.com>
When the break out or the WSL environment fails to start, the config
and connections should not be written. Placing them at the end of the
provisioning step will mitigate the issue.
[NO NEW TESTS NEEDED]
Signed-off-by: Gerard Braad <me@gbraad.nl>
- Allow creating sigstore signatures via --sign-by-sigstore-private-key .
Like existing --sign-by, it does not work remote (in this case
because we would have to copy the private key to the server).
- Allow passing a passphrase (which is mandatory for sigstore private keys)
via --sign-passphrase-file; if it is not provided, prompt interactively.
- Also, use that passphrase for --sign-by as well, allowing non-interactive
GPG use. (But --sign-passphrase-file can only be used with _one of_
--sign-by and --sign-by-sigstore-private-key.)
Note that unlike the existing code, (podman build) does not yet
implement sigstore (I'm not sure why it needs to, it seems not to
push images?) because Buildah does not expose the feature yet.
Also, (podman image sign) was not extended to support sigstore.
The test for this follows existing (podman image sign) tests
and doesn't work rootless; that could be improved by exposing
a registries.d override option.
The test for push is getting large; I didn't want to
start yet another registry container, but that would be an
alternative. In the future, Ginkgo's Ordered/BeforeAll
would allow starting a registry once and using it for two
tests.
Signed-off-by: Miloslav Trmač <mitr@redhat.com>
Before making / mutable/immutable, podman-machine checks if the mount is
being done in /home or /mnt. However the current check is always going
to be true:
```
!strings.HasPrefix(mount.Target, "/home") || !strings.HasPrefix(mount.Target, "/mnt")
```
is false when mount.Target starts with "/home" and mount.Target starts
with "/mnt", which cannot happen at the same time.
The correct check is:
```
!strings.HasPrefix(mount.Target, "/home") && !strings.HasPrefix(mount.Target, "/mnt")
```
which can also be written as:
```
!(strings.HasPrefix(mount.Target, "/home") || strings.HasPrefix(mount.Target, "/mnt"))
```
The impact is not too bad, it results in extra 'chattr -i' calls which
should be unneeded.
[NO NEW TESTS NEEDED]
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Option defaults in API must be the same as in CLI.
```
% podman image push --help
% podman image pull --help
% podman manifest push --help
% podman image search --help
```
All of these CLI commands them have --tls-verify=true by default:
```
--tls-verify require HTTPS and verify certificates when accessing the registry (default true)
```
As for `podman image build`, it doesn't have any means to control
`tlsVerify` parameter but it must be true by default.
Signed-off-by: Vladimir Kochnev <hashtable@yandex.ru>
Some refer to issues that are closed. Remove them.
Some are runc bugs that will never be fixed. Say so, and remove
the FIXME.
One (bps/iops) should probably be fixed. File an issue for it, and
update comment to include the issue# so my find-obsolete-skips script
can track it.
And one (rootless mount with a "kernel bug?" comment) is still
not fixed. Leave the skip, but add a comment documenting the symptom.
Signed-off-by: Ed Santiago <santiago@redhat.com>
