Makerpms: Error: invalid configuration

Bug
1

Since the last update on F31 this wk, I cannot build anymore with makerpms

makerpms *.spec
Error: invalid configuration, cannot specify resource limits without cgroups v2 and --cgroup-manager=systemd

I fixed the issue (or workaround ?) by adding in /usr/bin/makerpms

podman run \
    --hostname b$$.$(hostname) \
+    --cgroup-manager=systemd \
    --name "${container_id}" \
    --security-opt label=disable \

After this I can use makerpms and build the rpm

2 Likes
2

We have some informations


1 Like
3

Quoting the article, I guess the origin of the issue is here:

If you turn on cgroup v2 by default and replace cgroup v1, all of the container tools break.

The workaround could be erasing everything under $HOME/.local/share/containers/ (:warning: if you use podman containers just for makerpms…)

4

no it failed

Trying to pull registry.fedoraproject.org/nethserver/makerpms:7...
  manifest unknown: manifest unknown
Trying to pull registry.access.redhat.com/nethserver/makerpms:7...
  name unknown: Repo not found
Trying to pull registry.centos.org/nethserver/makerpms:7...
  manifest unknown: manifest unknown
Trying to pull docker.io/nethserver/makerpms:7...
Getting image source signatures
Copying blob ab5ef0e58194 done  
Copying blob 3821635d80bd done  
Copying blob 0ff2de6eda78 done  
Copying blob ff15eb414b83 done  
Copying blob a828d29c42e1 done  
Copying blob 147b8f435ee9 done  
Copying blob 0fc78a067faf done  
Copying config 3396314b6f done  
Writing manifest to image destination
Storing signatures
Error: invalid configuration, cannot specify resource limits without cgroups v2 and --cgroup-manager=systemd
5

Did you remove your patch?

6

of course

podman run \
    --hostname b$$.$(hostname) \
    --name "${container_id}" \
    --security-opt label=disable \
    --volume $PWD:/srv/makerpms/src:ro \
    --volume ${cachevol}:/var/cache/yum \
    --env NSVER \
    --env DIST \
    --tty \
    --interactive \
    ${image} \
    ${command:-makerpms} -s "$@"
7

As discussed in the issue tracker, the bug is not reproducible. If somebody else hits a similar problem we can continue this discussion.

Thanks to @stephdl for pointing it out!

1 Like
8

I had the same issue after update in FC31.
In my case the solution was to edit ~/.config/containers/libpod.conf and remove (or comment out) line
cgroup_manager = “cgroupfs”

3 Likes
9

systemd is default so it make sense @saki that if you remove the line it maked it workable

ping @davidep

1 Like
10

something new to add, after a new upgrade of Fedora 31 this wk, I have a new bug today

https://github.com/containers/libpod/issues/5513

in short the events_logfile_path = "" in ~/.config/containers/libpod.conf output some errors and I could see a new file .lock in my git files :slight_smile:

this is the error I had when I used makerpms : ERRO[0000] unable to write pod event: "open : no such file or directory"

this error is solved, you just need to comment it, but do you know what I found in this file @davidep

cgroup_manager = "cgroupfs" hence why I needed to set the cgroups to systemd in the /usr/bin/makerpms

1 Like