Resolution:
Seems it became mandatory to add parameter “–net=host”. At least once added it is working as before.
About a month ago I upgraded my system (archlinux) after which my (local) docker apps which require UI ceased working. During this upgrade docker got upgraded to version 1.10 (also kernel, if that matters). I was waiting a few weeks in anticipation of further upgrades to fix the issue, which is not the case. According to Google it appears I am the only one with this issue. The apps where working fine until the upgrade, e.g. below app was opening as separate window and I could work.
I could start docker apps until recently as follows:
kmymoney() {
xhost local: > /dev/null
docker_run=kmymoney
if [ ! -z "$1" ]; then
docker_run="$1"
fi
docker run --rm -ti \
-v /home/juser/tmp/dockerhome:/home/kmymoney/ \
-v /mnt/nas/data:/home/kmymoney/data \
-v /tmp/.X11-unix:/tmp/.X11-unix \
--device=/dev/snd:/dev/snd \
--device=/dev/dri:/dev/dri:rw \
-e DISPLAY=unix$DISPLAY \
kmymoney:2015-08-01 $docker_run
xhost -local: > /dev/null
}
Since the system (and docker) upgrade I am getting error:
cannot connect to X server unix:0
I can still use non-UI apps and I can also start UI apps as long as they are not using X server, i.e. command line only. For example, I can start above image with
“/bin/bash” for ‘$docker_run’ instead of “kmymoney”.
Rollback is not an option as docker images got migrated. I have also rebuild the image but still same issue.
I suspect I have done something incorrectly right from the beginning which might have been tolerated by system or Docker until the upgrade. Anyone having an idea?