Running mesos slave inside of a docker container and having working slave task recovery isn't supported at the moment. See: https://issues.apache.org/jira/browse/MESOS-2115
On Mon, May 18, 2015 at 4:47 AM, Grzegorz Graczyk <[email protected]> wrote: > 3-node cluster > CoreOS 675.0.0 > Mesos 0.22.1 > Marathon 0.8.2-RC2 > > Everything is run in containers, mesos slave run using command: > /usr/bin/docker run \ > --rm \ > --net=host \ > --pid=host \ > --name slave \ > -v /data/server/mesos-slave:/data/mesos-slave \ > -v /root/.dockercfg:/etc/.dockercfg \ > --privileged \ > -v /usr/lib64/libdevmapper.so.1.02:/usr/lib/libdevmapper.so.1.02 \ > -v /var/run/docker.sock:/var/run/docker.sock \ > -v /usr/bin/docker:/usr/local/bin/docker \ > -v /sys/fs/cgroup:/host/sys/fs/cgroup \ > -e GLOG_v=1 \ > mesosphere/mesos-slave:0.22.1-1.0.ubuntu1404 --containerizers=docker,mesos > --master=zk://`/get-zookeeper-peers.sh`/mesos > --hostname=private.`hostname` --ip=${ENS224_IPV4} > --resources=\"ports(*):[31000-32000]\" > --cgroups_hierarchy=/host/sys/fs/cgroup --work_dir=/data/mesos-slave > --logging_level=INFO > > After slave restart it succesfully re-registers in mesos master, then it > kills all tasks and starts them again. > Same happens when using mesos containerizer. > > Full logs: https://gist.github.com/gregory90/dd6930495fd655cf6691 > > Any help appreciated. >
