Don't use reexec for the rootlessport process, instead make it a
separate binary to reduce the memory usage. The problem with reexec is
that it will import all packages that podman uses and therefore loads a
lot of stuff into the heap. The rootlessport process however only needs
the rootlesskit library.
The memory usage is a concern since the rootlessport process will spawn
two process per container which has ports forwarded. The processes stay
until the container dies. On my laptop the current reexec version uses
47800 KB RSS. The new separate binary only uses 4540 KB RSS. This is
more than a 90% improvement.
The Makefile has been updated to compile the new binary and install it
to the libexec directory.
Fixes#10790
[NO TESTS NEEDED]
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Don't hardcode /usr/bin/podman in unit files: instead, use
template files with a path replaced at install time.
Because 'make' can be invoked repeatedly, with different
PREFIX, do not leave the generated files behind in our
work directory: wipe them immediately after install.
To get this to work, fix a longstanding bug in podman.spec.in,
a PREFIX that should've been DESTDIR.
Side note: #7023 made contrib/systemd/user a symlink
to .../system but did not update paths in Makefile.
The unrelated-looking path change you see here is
a belated correction for that.
Fixes: #10787
Signed-off-by: Ed Santiago <santiago@redhat.com>
Accept both "NO TESTS NEEDED" and "NO NEW TESTS NEEDED".
That was a usability mistake I made on Day One. Fixed it
in Buildah but oops never got around to fixing it here.
Also, fix the test suite script: remove a no-longer-working
test case (changelog.txt, removed in #11467) and add a new
test for commits that include the magic string.
Signed-off-by: Ed Santiago <santiago@redhat.com>
Original workaround https://github.com/containers/podman/pull/11821
During VM image build, a number of packages are downloaded but not
installed, since they may interfere with some testing. Then at runtime,
where required, the packages are installed from cache and used.
However, between image build and runtime it's possible the repository
contents change, which will invalidate the package cache. Since the
`--no-download --ignore-missing` options were used, the install will
fail.
Ref: https://github.com/containers/automation_images/issues/95
Fortunately, when it comes to the docker packages, no other dependencies
are required and so `apt-get` isn't required. Switch to using a simple
dpkg install command on the necessary files. If this ever breaks due
to new dependencies, the list of files may simply be updated.
Signed-off-by: Chris Evich <cevich@redhat.com>
It looks like the containerd.io package is not present anymore in the
package cache which ultimately breaks CI since it's a requirement for
docker.
Hence, download the few packages instead of relying on the cache.
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
Hi there! I noticed that the MSI file built for Windows was building in 32-bit mode, even though Podman is 64-bit only. I added a flag to wixl in the Makefile to tell it to build a 64-bit MSI, and I adjusted the podman.wxs file to use the right settings for 64-bit programs.
Signed-off-by: GitHub <noreply@github.com>
Add execution of the downstream gitlab-runner tests using
rootless podman through the magic of socket-level
docker compatibility. Include a comment suggesting how
to temporarily disable the test in case it fails beyond
podman code scope.
Signed-off-by: Chris Evich <cevich@redhat.com>
Add a link to the podman images readme.md to Dan's recent post
on Enable Sysadm about running containers inside of Podman
Fixes: https://github.com/containers/buildah/issues/3119
Signed-off-by: TomSweeneyRedHat <tsweeney@redhat.com>
Fixes#11417
Cross-building the podman-remote documentation requires a functional
native architecture executable. However `make` only deals with
files/timestamps, it doesn't understand if an existing binary will
function on the system or not. This makes building cross-platform
releases incredibly accident-prone and fragile.
A practical way to deal with this, is via multiple conditional (nested)
`make` calls along with careful manipulation of `$GOOS` and `$GOARCH`.
Also, when cross-building releases be kind to humans and cleanup
any non-native binaries left behind.
Update the `Alt Arch. Cross` Cirrus-CI task to build release archives
for all Linux architectures supported by golang and podman. Update
the `OSX Cross` task to additionally build for the M1 (arm64)
architecture.
Finally, update the release process documentation to reflect the
new locations (Cirrus-CI task names) for the release archives. Include
a note about additional manual work being required to produce the
signed `.dmg` file for MacOS.
Signed-off-by: Chris Evich <cevich@redhat.com>
Rootless cni with ipv6 needs the `ip6_tables` module loaded, normally
the cni plugins will load this module but as rootless it does not have
the necessary permission to do so. Therefore we load it manually.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
The changelog.txt file hasn't been kept in sync with release tags,
especially on main, so remove it.
The release notes will be featured in RELEASE_NOTES.md.
Signed-off-by: jesperpedersen <jesper.pedersen@redhat.com>
[NO TESTS NEEDED]
This becomes a problem on hosts with upgraded policies. Ref:
https://github.com/containers/podman/issues/10522
Also, made a small change to compose-test setup to reduce runtime.
Signed-off-by: Chris Evich <cevich@redhat.com>
The automation workflow was altered in recent history to build images
daily, even if the podman version didn't change. This was is necessary
so that any updates/security vulnerabilities in ancillary packages are
incorporated quickly. However, documentation was never updated to
reflect this change. This commit puts the two in sync.
Signed-off-by: Chris Evich <cevich@redhat.com>
Significant bitrot results in almost immediate test failure. This
commit adds only the very basic, bare-minimum needed to get them
started.
***TESTING RESULTS ARE IGNORED***
Signed-off-by: Chris Evich <cevich@redhat.com>
There seems to be a bug in rpm, where it fails silently if you specify
rpm --restore --quiet shadow-utils.
rpm --restore shadow-utils 2> /dev/null
Does the right thing.
[NO TESTS NEEDED] Might add tests from buildah, once we have them
working correctly.
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
Rather then reinstalling shadow-utils to fix permissions,
just restore the correct permissions.
[NO TESTS NEEDED] Since this does not affect Podman, just the prebuilt
images on quay.io/podman.
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
Require the network to be online in all (generated) systemd units to
make sure that containers and Podman run only after the network has been
fully configured.
Fixes: #10655
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
This makes sure, that the podman auto-update is not executed exactly at midnight for the same time always.
If many things do the same and many services use this keyword and react at midnight, this can cause a lot of stress to a server.
Thus, this adds a 900s/15min delay.
As [the arch wiki says](https://wiki.archlinux.org/title/Systemd/Timers#Realtime_timer):
> Special event expressions like daily and weekly refer to specific start times and thus any timers sharing such calendar events will start simultaneously. Timers sharing start events can cause poor system performance if the timers' services compete for system resources. The RandomizedDelaySec option in the [Timer] section avoids this problem by randomly staggering the start time of each timer. See systemd.timer(5).
Signed-off-by: rugk <rugk+git@posteo.de>
The rootless cni namespace needs a valid /etc/resolv.conf file. On some
distros is a symlink to somewhere under /run. Because the kernel will
follow the symlink before mounting, it is not possible to mount a file
at exactly /etc/resolv.conf. We have to ensure that the link target will
be available in the rootless cni mount ns.
Fixes#10855
Also fixed a bug in the /var/lib/cni directory lookup logic. It used
`filepath.Base` instead of `filepath.Dir` and thus looping infinitely.
Fixes#10857
[NO TESTS NEEDED]
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
The new cobra v1.2.0 release brings a number of bug fixes for shell
completion scripts. Regenerate the scripts with `make completions`
to sync them with the upstream version, currently we have some custom
ones to avoid some upstream bugs. Because the new cobra version has
all fixes we should use the upstream scripts.
Add a check to CI to ensure we always use the up to date scripts.
[NO TESTS NEEDED]
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
We no longer need to install /etc/cni/net.d/87-podman-bridge.conflist
so install.cni isn't needed either.
Signed-off-by: Lokesh Mandvekar <lsm5@fedoraproject.org>
The systemd-tmpfiles configuration is meant preserve important
paths in /tmp that are used by Podman against deletion by
systemd. However, not all paths we previously used were included.
Some older versions used the `/tmp/use-$UID/libpod` directory
instead (when `/run/user/$UID` was unavailable).
Add an entry for these old paths to ensure tmpfiles treats the
directory correctly.
Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1960948
Signed-off-by: Matthew Heon <mheon@redhat.com>
Docker VOLUMEs will inherit permissions from an existing directory at the same
path. If the path does not exist, the directory will be owned by root which
makes this image unusable in rootless mode.
Signed-off-by: Blake Burkhart <blake.burkhart@us.af.mil>
* Add podman-restart systemd unit file and add it to podman RPM package
* Fix podman start to filter all containers + unit test
Signed-off-by: Boaz Shuster <boaz.shuster.github@gmail.com>
Extend the systemd auto-update service to prune images after an update
has run. As reported by a user [1], auto updates can over time cause
the disk to run out of space. With Edge being a target use case, we
need to make sure that systems can run without much supervision, so
let's make sure to run `podman image prune` to clean up dangling images.
[1] https://twitter.com/r_isc_y/status/1388981737011793921Fixes: #10190
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
Commit 7f2c27d43f added an invalid value for the log_driver in the
containers.conf file inside the podman image.
Fixes#10312
Signed-off-by: Paul Holzinger <paul.holzinger@web.de>
Update the order of image documentation to be from most to least stable.
Similarly, avoid depending on execution of upstream podman, when
building/pushing. It's easily possible for this build to function but
execution to fail due to some partially implemented feature.
Also, ensure images tagged `latest` are pushed for every matrix
item. For 'upstream' and 'testing', this replaces use of the
'master' tag.
Lastly, update workflow comments and split the 'podman' and 'containers'
FQIN steps and outputs to improve readability.
Signed-off-by: Chris Evich <cevich@redhat.com>
Also simplify `lib.sh` after supporting changes incorporated
into automation library 2.x+ (present in all VM and container images).
* No need to force-load `/etc/profile` and handle it's expectation
to **not** being in `errexit` mode.
* Slightly re-arrange loading of automation library files for
clarity.
* Update comments.
Signed-off-by: Chris Evich <cevich@redhat.com>
Distro builds on Fedora and Kubic projects use GO111MODULE=off
by default which are currently failing. This commit fixes it and
going forward, podman CI will also indicate failures in rpm builds.
The additional LDFLAGS have been removed from the spec file
which is not ideal. But, currently we only use the spec file
to check if the rpm builds fine. We can fix the LDFLAGS in a
later commit when we're working on packit integration.
conmon build has also been removed from podman.spec.in because the COPR
for which it was provided has been discontinued.
[NO TESTS NEEDED]
Fixes: #10009
Signed-off-by: Lokesh Mandvekar <lsm5@fedoraproject.org>
docker-compose test continues to flake even after #9961.
Let's try to get some useful data from the failures, by:
* adding -S (--show-error) to curl. With just -s (--silent),
curl is completely quiet. With -S, it displays errors.
(Not in TAP form, but I'm OK with that)
* oops, adding safety checks to the fix from #9961 (it
was inadvertently clobbering the curl exit status)
And, as long as I'm in this code: logformatter was not
highlighting these results, because the '1..N' TAP line
needs to be spit out at the end. Have test-compose emit
a 'TAP' header <http://testanything.org/> and make
logformatter recognize it.
Signed-off-by: Ed Santiago <santiago@redhat.com>
* Incorporate changes from abandoned #9918: Use dedicated `bin`
sub-directories for `windows` and `darwin` when building
`podman-remote`. The linux flavor remains under `bin` as before.
* Fix MacOS Documentation-generation for release-packaging.
The `install-podman-remote-%-docs` target requires local execution
of `podman-remote`, but it was assuming GOOS=linux. Fix this
by dynamically discovering the local OS/architecture type while
still permitting cross-building of MacOS binaries under Linux.
* Unify temporary directory/file behavior to use a common template.
In case of left-over temporary items left in the repository,
update the `clean` target accordingly to remove them.
* Fix broken podman-remote-static and MacOS release archive targets
mismatching the `podman-remote-%` target. Disambiguate this target
for all platforms by spelling each out in full, instead of using
a wild-card recipe.
* Fix Windows-installer target to properly recognize existing
output files and not constantly rebuild every time.
* Include the podman version number in the Windows-installer target
in case a user downloads multiple releases.
* Include a subdirectory containing the podman version number for
both `tar.gz` and `zip` targets. This prevents users clobbering
existing directories when un-archiving from releases.
Signed-off-by: Chris Evich <cevich@redhat.com>
Easton Man
Paul Holzinger
Ed Santiago
OpenShift Merge Robot
Matthew Mosesohn
Chris Evich
Valentin Rothberg
Easton Pillay
TomSweeneyRedHat
Daniel J Walsh
jesperpedersen
openshift-ci[bot]
rugk
Lokesh Mandvekar
Matthew Heon
Blake Burkhart
Boaz Shuster
Paul Holzinger