The remote client (podman --remote) was incorrectly throwing an error
when --detach-keys="" was specified for attach, run, or start commands.
According to documentation and the v1.7.0 release notes, specifying an
empty string should disable detaching, not cause an error.
Fixes: #27414
Signed-off-by: shiavm006 <shivammittal42006@gmail.com>
This variable is dead code as far as I can tell. I think it
got cargo culted from something similar in skopeo, where it *is*
used:
85598438ce/Makefile (L75)
etc.
(Instead it looks like there's a `PODMANCMD` here)
But I'm effectively using this PR as a way to suggest aligning
with what we're doing in bootc, where we have a core principle
that `Makefile` should *never* itself spawn containers (or VMs etc).
All the rules in Makefile are things that should work when e.g.
building RPMs or debs or the like.
Those tasks are things that are done via `Justfile`:
https://github.com/bootc-dev/bootc/blob/main/Justfile
So for example to align, we'd add a `Justfile` here and then
`make validatepr` would move to `just validatepr`.
Signed-off-by: Colin Walters <walters@verbum.org>
Perhaps exposing some our critical pull requests requirements in a checklist might help users. As of now, we do this as a "soft open".
Co-authored-by: Jan Rodák <hony.com@seznam.cz>
Co-authored-by: Ashley Cui <ashleycui16@gmail.com>
Signed-off-by: Brent Baude <bbaude@redhat.com>
A condition was changed in the refgactor of init where duplicate names would be allowed but no machine was created. Duplicate names are not permitted and should return an error.
Signed-off-by: Brent Baude <bbaude@redhat.com>
The windows installer doesn't like -rc's because it's not strictly semver, but this error can be suppressed.
Signed-off-by: Ashley Cui <acui@redhat.com>
The WSL tests for `podman machine` are flaking with unreasonable
frequency right now. Last Friday, I needed to re-run a job 8
times to get it to pass and complete the release of 5.7 RC1. This
is simply not sustainable. Mark the tests allowed-failure until
this can be resolved.
Signed-off-by: Matt Heon <mheon@redhat.com>
Add the ability for users to override the default provider when creating mahcines. The new flag is `--provider` and allows you to specifiy a valid vmtype for the platform. This PR also removes the previous list test where we tested listing all providers. I added a PR for testing --provider which includes a standard `machine ls` which defaults now to showing all providers.
Signed-off-by: Brent Baude <bbaude@redhat.com>
Bump bundled krunkit to 1.1.0. For podman, this release means
better I/O performance due to this libkrun version defaulting to
a more relaxed disk image sync on macOS, and raising the vCPU
limit to the maximum allowed by Hypervisor.framework.
Fixes: #27216
Signed-off-by: Sergio Lopez <slp@redhat.com>
This also includes a number of significant changes to the SQLite
state made possible by removal of the legacy DB.
1. Enable database unit tests for SQLite state, with numerous
tweaks to get tests passing. Most notable changes are to
container removal - where we previously didn't return an error
if there was no container to remove - and RemovePodContainers,
which I don't think ever worked properly from my reading of
the failures.
2. Removal of AddContainerToPod/RemoveContainerToPod. On SQLite,
these functions are identical to AddContainer/RemoveContainer
and there is no reason to retain duplicates.
3. Removal of SafeRewriteContainerConfig - it's identical to
RewriteContainerConfig in SQLite, no reason to have duplicate
entrypoints.
As an exciting side-note, this removes Podman's requirement that
containers and pods cannot share a name, which was a BoltDB
restriction only.
Signed-off-by: Matt Heon <matthew.heon@pm.me>
For Podman 6, we still have providers and will continue to have a default provider for each platform. But where a platform has multiple providers, we want users to be able to cross provider boudnaries imposed in Podman 4/5. The key change is to look up virtual machines by name, as before, but to then also iterate all possible providers. As of this PR, init will still only create with the default provider, but a subsequent PR will introdouce an provider override.
I also removed the "--all-providers" command line option on `podman
machine ls` because it no longer makes sense. And I marked the all
provider list test to be skipped.
Signed-off-by: Brent Baude <bbaude@redhat.com>
Jan does great contributions, reviews other PRs and helps with issue
triage for a while now. He definitely deserves to be a full Maintainer
for Podman so he can properly merge PRs.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
