shouldn't the current directory be the default context for "podman build"?
by Robert P. J. Day
"man podman-build" suggests that the context argument is optional:
SYNOPSIS
podman build [options] [context]
podman image build [options] [context]
...
If no context directory is specified, then Podman will assume
the current working directory as the build context, which
should contain the Containerfile.
but if i have a directory with nothing but a Containerfile, i get:
$ podman build
Error: no context directory specified, and no containerfile specified
$
OTOH, specifying context of current directory:
$ podman build .
STEP 1: FROM alpine:latest
... etc etc ...
thoughts?
rday
1 week, 4 days
mqueue msg_max in rootless container
by Michael Ivanov
Hallo!
I'm trying to run my application in podman rootless container and I stumble
on following problem: my program needs /proc/sys/fs/mqueue/msg_max to be at
least 256, but in running container this value is just 10. When I try to
specify this parameter while running the image (--sysctl 'fs.mqueue.msg_max=256')
I get the following error:
Error: open /proc/sys/fs/mqueue/msg_max: Permission denied: OCI permission denied
and container is not created.
My host where container is being run has this parameter set to 256. How can I
expose current host setting for msg_max to my container?
Best regards,
--
\ / | |
(OvO) | Михаил Иванов |
(^^^) | |
\^/ | E-mail: ivans(a)isle.spb.ru |
^ ^ | |
1 year
Python process unable to write files
by Josh Berkus
All,
I'm porting a legacy app to containers, and having an issue where
apparently it can't write files while running in podman.
Is there any reason why a python process, or child process, running as
container-root would be unable to write to either the ephemeral
filesystem of the container, or to mounted volumes?
Basically, here's the situation:
- pyhton app with many child processes
- all of them run as container-root
- app is supposed to write logs to files (yes, I know)
- app does not write any logs to any files; in fact, the log-dir
initialization appears to fail (no error messages, though, because it's
not logging)
- one other process which is supposed to write cache to a dir does not do so
- all of these directories are under /app/ a directory COPYd into the
image definition, not /var/ or home
- have tried both with these dirs as local to the container, and as
mounted volumes on the host system
- if I exec in to the container as container-root, I can write files to
those dirs
- SELinux denial log on the host does not show any denials
It is entirely possible that this is a problem with the legacy app and
is not a podman thing at all. I'm asking here because I want to
eliminate podman as a potential cause of the problem.
--
-- Josh Berkus
Kubernetes Community Architect
OSPO, OCTO
3 years, 4 months
Why 'podman-docker' package conflicts with 'docker'
by Nishant Nayan
Hi,
I have a question, why does 'podman-docker' package conflicts with
docker ?
I tried installing both at the same time but faced conflict.
Is the conflict is by design?
I have seen specification of podman-docker in the following link
https://archlinux.org/packages/community/x86_64/podman-docker/
It says it conflicts with docker, but I want to know the reason for such
spec.
I would like to know why 'podman-docker' and 'docker' can't coexist?
It would be really helpful.
Regards
Nishant Nayan
3 years, 4 months
podman slirp4netns IPv6
by Hendrik Haddorp
Hi,
I'm using podman 2.2.1 rootless with slirp4netns and IPv6 enabled
(--network slirp4netns:enable_ipv6=true). Straight after starting my
container the code inside tries to connect to a server using IPv6 but
fails due to the network being unreachable. However if I add a small
delay before trying to connect it works. If I use IPv4 this seems to
work right after the container is started, which of course still takes a
few ms until the code is started.
I can actually easily recreate the using podman 3.1.2 as well.
this works:
podman run -it --network slirp4netns:enable_ipv6=true --entrypoint bash
fedora:34 -c "curl -v -4 https://www.google.com"
this fails:
podman run -it --network slirp4netns:enable_ipv6=true --entrypoint bash
fedora:34 -c "curl -v -6 https://www.google.com"
but works with a delay:
podman run -it --network slirp4netns:enable_ipv6=true --entrypoint bash
fedora:34 -c "sleep 2; curl -v -6 https://www.google.com"
Is the network setup done after the container is already started and
thus this needs to be expected?
thanks,
Hendrik
3 years, 5 months
podman images | head crashes
by Manvendra Bhangui
Not a big deal. podman crashes when output is fed to any program that
closes input before reading the full output from podman
Maybe SIGPIPE needs to be handled
$ podman images|head
REPOSITORY TAG IMAGE ID
CREATED SIZE
localhost/tinydnssec alpine d28977cc2347 40
minutes ago 98.5 MB
localhost/indimail alpine 9c520b04ccda 2 days
ago 547 MB
localhost/indimail-mta alpine 3ced90144c5b 3 days
ago 252 MB
localhost/indimail-mta tumbleweed 999a86c2bc61 12
days ago 413 MB
localhost/indimail-mta leap15.3 899596f466a1 12
days ago 500 MB
localhost/indimail-mta debian10 f3a8194282d7 3
weeks ago 306 MB
registry.opensuse.org/opensuse/leap 15.3 accc3d285fe7 3
weeks ago 108 MB
registry.opensuse.org/opensuse/leap latest accc3d285fe7 3
weeks ago 108 MB
docker.io/library/almalinux 8 7a497d63e726 3
weeks ago 216 MB
signal 13 received but handler not on signal stack
fatal error: non-Go code set up signal handler without SA_ONSTACK flag
runtime stack:
runtime.throw(0x558c5d761f18, 0x39)
/usr/lib/golang/src/runtime/panic.go:1117 +0x74 fp=0xc000658cb8
sp=0xc000658c88 pc=0x558c5c3a9d34
runtime.sigNotOnStack(0xd)
/usr/lib/golang/src/runtime/signal_unix.go:918 +0x85
fp=0xc000658cd8 sp=0xc000658cb8 pc=0x558c5c3c1e65
runtime.adjustSignalStack(0xc00000000d, 0xc000184400, 0xc000658d58,
0x558c5dc4bf00)
/usr/lib/golang/src/runtime/signal_unix.go:509 +0x285
fp=0xc000658d30 sp=0xc000658cd8 pc=0x558c5c3c0d05
runtime.sigtrampgo(0xd, 0xc000658f30, 0xc000658e00)
/usr/lib/golang/src/runtime/signal_unix.go:449 +0x13f
fp=0xc000658da8 sp=0xc000658d30 pc=0x558c5c3c097f
runtime: unexpected return pc for runtime.sigtramp called from
0x7f08d4cf1320
stack: frame={sp:0xc000658da8, fp:0xc000658e00}
stack=[0xc000650cd8,0xc0006590d8)
000000c000658ca8: 000000c000658cc8 0000558c5c3c1e65
<runtime.sigNotOnStack+133>
000000c000658cb8: 0000558c5d761f18 0000000000000039
Full output here
https://termbin.com/x7gr
--
Regards Manvendra - http://www.indimail.org
GPG Pub Key
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xC7CBC760014D250C
3 years, 5 months
What is podman-plugins?
by fugkco
Hi,
I'm just looking through my system and I'm curious what the packages podman-plugin and podman-machine-cni are. I have them both installed but the only thing they contain is some docs.
$ dpkg -S podman-machine-cni podman-plugins
podman-machine-cni: /usr/share/doc/podman-machine-cni/changelog.Debian.gz
podman-machine-cni: /usr/share/doc/podman-machine-cni
podman-machine-cni: /usr/share/doc/podman-machine-cni/copyright
podman-plugins: /usr/share/doc/podman-plugins/changelog.Debian.gz
podman-plugins: /usr/share/doc/podman-plugins
podman-plugins: /usr/share/doc/podman-plugins/copyright
More importantly, I'm currently scripting some script that auto compiles podman et al (reason being the Kubic repos tend to be a little bit behind), and trying to figure out what it is that I would need to install for those two packages (if at all).
Thanks
3 years, 5 months
named pipes doing a disappearing act
by Manvendra Bhangui
I have created images for my software using the podman build command. My
application uses named pipes to communicate. So during installation I
create the fifo using the mkfifo command. But when I start the container
I find all named pipes missing. Every time I start the container,
I have to re-create the pipes using mkfifo. I have tried images from
fedora, debian, ubuntu, opensuse and the problem is being faced on
all. Apart from named pipes, other types of files are getting saved.
As a workaround, I run a script using docker-entrypoint to recreate
the named pipes.
This is not happening if I use docker command to run the image.
Even images built by the docker build command, when I use
podman to run the image, the named pipes go missing.
This is all I did to simulate the issue
1 podman pull someimage
2 podman run --name test -ti imageid bash
cd /root
mkfifo test
3 podman save test test
4 podman stop test
5 podman run --name test -ti newimageid bash
cd /root
ls -l # this shows that the file is missing and didn't get saved
I have pasted the result of the test using podman and docker here
https://termbin.com/7s5u
Is there anything special I need to do to have these files seen and accessible?
--
Regards Manvendra - http://www.indimail.org
GPG Pub Key
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xC7CBC760014D250C
3 years, 5 months
Looking for Topics - Podman Community Meeting August 5, 2021 11:00 a.m. EDT (UTC-4)
by Tom Sweeney
Hi All,
The next Podman Community Meeting is just under two weeks away and at
the moment I do not have a single topic. If there's something that
you'd like to present, or a topic that you'd like one of the maintainers
talk more about, please reply to this email. Remember, topics don't
have to be just about Podman. We're happy to consider topics from
anything related to containers, including your own project, especially
if they have a tie with Buildah, Podman or Skopeo.
Thanks!
t
3 years, 5 months
Recommended way to manage events.log file
by Dale Baley
Hi, we rely on podman events via file for our workflows. Is there a recommended way to truncate/rotate/move the events.log file without losing potential event logs while doing so? journald isn't an option.
Thanks in advance
3 years, 5 months