ActiveState=failed after running "systemctl --user start podman.service" #15927
Labels
kind/bug
Categorizes issue or PR as related to a bug.
locked - please file new issue/PR
Assist humans wanting to comment on an old issue or PR with locked comments.
Comments
|
If I create the new user like this then everything works for both alternatives
I wondered if this issue is a general problem related to socket activation and not having the user namespace set up, so I tried out socket activation of containers.
Interestingly the command I tried the same echo test again but now with I tried the same echo test again but now with Conclusion: This issue seems related to socket activation and not having the user namespace set up. |
|
New test with Fedora CoreOS. Podman version 4.2.1. test: "podman system service" (via "systemctl --user start podman.service")test.shtest_unshare.shIs almost the same as test.sh but with an additional line --- /tmp/test.sh 2022-09-25 08:41:05.148456553 +0000
+++ /tmp/test_unshare.sh 2022-09-25 08:40:38.141553130 +0000
@@ -11,6 +11,7 @@
set -o nounset
systemctl --user start podman.socket
+podman unshare /bin/true
systemctl --user start podman.service
sleep $seconds
systemctl --user show -P ActiveState podman.serviceWhen using a sleep-value of 0 seconds, the service is active. I guess the service has not yet failed. test: socket activation of containerssocketactivation.shsocketactivation_unshare.shIs almost the same as socketactivation.sh but with an additional line [root@asus ~]# diff -u /tmp/socketactivation.sh /tmp/socketactivation_unshare.sh
--- /tmp/socketactivation.sh 2022-09-28 21:00:48.704516815 +0200
+++ /tmp/socketactivation_unshare.sh 2022-09-28 21:01:00.028254794 +0200
@@ -11,6 +11,7 @@
set -o nounset
systemctl --user start podman.socket
+podman unshare /bin/true
systemd-socket-activate -l /home/${username}/sock podman run --rm --name echo --network=none ghcr.io/eriksjolund/socket-activate-echo > ~/stdout 2> ~/stderr &
sleep 1
echo hello | socat -t 100 - unix:/home/${username}/sockThe reply hello can not be seen when running socketactivation.sh: The reply hello can be seen when running socketactivation_unshare.sh: About the system |
This was referenced Oct 15, 2022
eriksjolund
added a commit
to eriksjolund/podman
that referenced
this issue
Oct 15, 2022
[NO NEW TESTS NEEDED] Fixes: containers#15927 Signed-off-by: Erik Sjölund <[email protected]>
mheon
pushed a commit
to mheon/libpod
that referenced
this issue
Oct 18, 2022
[NO NEW TESTS NEEDED] Fixes: containers#15927 Signed-off-by: Erik Sjölund <[email protected]>
github-actions
bot
added
the
locked - please file new issue/PR
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Is this a BUG REPORT or FEATURE REQUEST? (leave only one on its own line)
/kind bug
Description
The systemd user service podman.service is failed after running
systemctl --user start podman.service.(I think this explicit starting of podman.service with socket activation should be supported although it is not the standard way of how to start podman.service).
A sidenote: Interestingly even when starting podman.service the standard way (i.e. a client connects to podman.socket), the log shows
and
and
Steps to reproduce the issue:
Create a new user and log in to it
Start the systemd user service podman.service with socket activation. Note, often socket-activated services
are started by systemd when the first client connects. In this case systemd starts podman.service with socket activation without waiting for the first client connection.
Run
systemctl --user show -P ActiveState podman.serviceCheck the podman.service log
Describe the results you received:
At step 3:
Describe the results you expected:
Either
or
The service should be active at first but then becoming inactive after the inactivity timeout (that can be set with the
podman system serviceoption --time)Additional information you deem important (e.g. issue happens only occasionally):
If I start podman.service (the standard way) by running
instead of running
I also see the error messages
Error: failed to start API service: accept unixgram:and
podman.service: Failed with result 'exit-code'.andThe unit UNIT has entered the 'failed' state with result 'exit-code'.See the log
If I run
systemctl --user show -P ActiveState podman.serviceafter the curl command, it first shows active and after some seconds inactive:Output of
podman version:Output of
podman info:Package info (e.g. output of
rpm -q podmanorapt list podman):The text was updated successfully, but these errors were encountered: