@Gnouc It's probably best to open a bug against OpenShift Origin and close this one:
https://github.com/openshift/origin/issues

@dims why Openshift Origin?

If I ran kubernetes alone, the same issue occur.

I think we should state this an OS bug, or make kubelet handle this situation

@smarterclayton - what's the best venue for this bug?

This is a Kubernetes bug - the line referenced by @Gnouc points out that the kubelet assumes the docker binary is called "docker" which is not necessarily the case.

I think at a minimum that should not be a constant, but be defaulted. The workaround for now is to hardlink the docker-current process you have to docker.

@smarterclayton thanks, sounds good. I have heard many times that there are patches/hardening to docker and kubernetes and was not sure.

@smarterclayton If you don't mind, I can push the fix for this issue.

@smarterclayton I'm having the same issue. If you don't mind explaining how to hardlink docker-current process to docker. Thanks

ln docker-current docker

thanks

I couldn't do a hard link in a bin directory as there already is a file named docker, so i moved docker-current, renamed it to docker and edited the /usr/lib/systemd/system/docker.service file with the new location of the renamed docker-current file.
Now I don't get the error, but get ������ in browser when I try to access the console.

My didn't access it with https.

@derekwaynecarr system container Cgrouping is broken with the latest docker packages due to the process name changing.

@acavas1 : can you try setting the pidfile parameter for your docker daemon? that may be an easier work around?

--pidfile=/var/run/docker.pid

Thanks,
Dims

@acavas1 Change the docker script to use #!/bin/bash, then replace the exec line with exec -a docker.

This has been fixed in Kubernetes here:
#25907

The origin cherry pick is here: openshift/origin#9060

@TomasTomecek @smarterclayton : i am also facing the same issue in origin 1.2.0 with quick installation method. Can anybody tell me is this resolved and whats docker-current. how to use the workaround.

Hi @priyanka5, let's move this discussion to the origin repository (issue openshift/origin#9060)

@dims I tried your solution on origin 1.2.0 (--pidfile in /etc/sysconfig/docker and restart) but this doesn't seem to help for me

@Gnouc How does your /usr/bin/docker looks like? Tried to configure your changes but still the log after restarting
Still same issue on

oc v1.2.0
kubernetes v1.2.0-36-g4a3f9c5

@lvthillo Here is it:

#!/bin/sh
. /etc/sysconfig/docker
[ -e "${DOCKERBINARY}" ] || DOCKERBINARY=/usr/bin/docker-current
if [ ! -f /usr/bin/docker-current ]; then
    DOCKERBINARY=/usr/bin/docker-latest
fi
if [[ ${DOCKERBINARY} != "/usr/bin/docker-current" && ${DOCKERBINARY} != /usr/bin/docker-latest ]]; then
    echo "DOCKERBINARY has been set to an invalid value:" $DOCKERBINARY
    echo ""
    echo "Please set DOCKERBINARY to /usr/bin/docker-current or /usr/bin/docker-latest
by editing /etc/sysconfig/docker"
else
    exec ${DOCKERBINARY} "$@"

Change the shebang to #!/bin/bash, then exec ${DOCKERBINARY} "$@" to exec -a docker ${DOCKERBINARY} "$@"

@Gnouc Thanks for the fast reply.
Now my /usr/bin/docker looks like this:

#!/bin/bash
. /etc/sysconfig/docker
[ -e "${DOCKERBINARY}" ] || DOCKERBINARY=/usr/bin/docker-current
if [ ! -f /usr/bin/docker-current ]; then
    DOCKERBINARY=/usr/bin/docker-latest
fi
if [[ ${DOCKERBINARY} != "/usr/bin/docker-current" && ${DOCKERBINARY} != /usr/bin/docker-latest ]]; then
    echo "DOCKERBINARY has been set to an invalid value:" $DOCKERBINARY
    echo ""
    echo "Please set DOCKERBINARY to /usr/bin/docker-current or /usr/bin/docker-latest
by editing /etc/sysconfig/docker"
else
    exec -a docker ${DOCKERBINARY} "$@"
fi

I restarted everything:

sudo service docker restart
sudo service origin-node restart

in logs:
E0629 10:42:53.022798 9519 container_manager_linux.go:267] failed to detect process id for "docker" - failed to find pid of "docker": exit status 1

What I don't really understand is the fact that:

pidof docker returns nothing
pidof docker-current returns pid
cat /var/run/docker.pid returns same pid as docker-current.

"/usr/bin/docker" is not used from the systemd.
See /usr/lib/systemd/system/docker.service

@khanamura
What should I edit?

[Unit]
Description=Docker Application Container Engine
Documentation=http://docs.docker.com
After=network.target
Wants=docker-storage-setup.service

[Service]
Type=notify
NotifyAccess=all
EnvironmentFile=-/etc/sysconfig/docker
EnvironmentFile=-/etc/sysconfig/docker-storage
EnvironmentFile=-/etc/sysconfig/docker-network
Environment=GOTRACEBACK=crash
ExecStart=/bin/sh -c '/usr/bin/docker-current daemon \
          --exec-opt native.cgroupdriver=systemd \
          $OPTIONS \
          $DOCKER_STORAGE_OPTIONS \
          $DOCKER_NETWORK_OPTIONS \
          $ADD_REGISTRY \
          $BLOCK_REGISTRY \
          $INSECURE_REGISTRY \
          2>&1 | /usr/bin/forward-journald -tag docker'
LimitNOFILE=1048576
LimitNPROC=1048576
LimitCORE=infinity
TimeoutStartSec=0
MountFlags=slave
Restart=on-abnormal
StandardOutput=null
StandardError=null

[Install]
WantedBy=multi-user.target

@lvthillo Change sh -c '/usr/bin/docker-current ... to sh -c 'exec -a /usr/bin/docker-current ...

@lvthillo
Change
sh -c '/usr/bin/docker-current ...
to
sh -c 'exec -a docker /usr/bin/docker-current ...

and then

systemctl daemon-reload
systemctl restart docker.service
systemctl restart origin-node.service

@khanamura @Gnouc Thanks, it works

@khanamura
Hi sir,
i tried to execute the some steps above. exactly the same by some time it cgoup changes to docker and some time still docker-current.
i am fed up with this issue that takes from me too much time.
please help