Thanks, you saved my life :) Exactly, I am running on CoreOS and I am going to write the unit files needed to make it run.
To lock the masters and the zookeeper daemons to the same nodes I am considering adding a meta tag to the master nodes, moreover I think publishing their hostnames to etcd so that no manual configuration will be needed. I think that this can be the effortless way to build a Mesos cluster. I will give you the repository name once I will make it work ;) @Kashif The containers will run on the host node and not inside the slave container > Le 27 oct. 2014 à 20:13, Grzegorz Graczyk <[email protected]> a écrit : > > Change name of container for something else, it can't start with "mesos-" > because that's how mesos knows which containers are managed by mesos. It just > kills itself. > Are you running this on CoreOS? > Read this thread: > https://www.mail-archive.com/[email protected]/msg01602.html > <https://www.mail-archive.com/[email protected]/msg01602.html> > > On 27 October 2014 18:33, Alessandro Siragusa <[email protected] > <mailto:[email protected]>> wrote: > Hi all, > > I’m trying to run mesos-slave in a Docker container with the containerizers > docker,mesos : > > $ docker run -ti --rm --name=mesos-slave --net=host -e > MESOS_ISOLATOR=cgroups/cpu,cgroups/mem -e MESOS_CONTAINERIZERS=docker,mesos > -e MESOS_HOSTNAME=slave-hostname -e MESOS_IP=1.2.3.4 -v > /run/docker.sock:/run/docker.sock -v /sys:/sys -e MESOS_LOG_DIR=/var/log -e > MESOS_MASTER=zk://ZOOKEEPER_IPS:2181/mesos <> -p 5051:5051 -e > MESOS_CHECKPOINT=false mesos-slave > I1027 16:53:04.851111 1 logging.cpp:142] INFO level logging started! > I1027 16:53:04.851359 1 main.cpp:126] Build: 2014-09-23 05:36:09 by root > I1027 16:53:04.851375 1 main.cpp:128] Version: 0.20.1 > I1027 16:53:04.851384 1 main.cpp:131] Git tag: 0.20.1 > I1027 16:53:04.851393 1 main.cpp:135] Git SHA: > fe0a39112f3304283f970f1b08b322b1e970829d > I1027 16:53:05.853960 1 containerizer.cpp:89] Using isolation: > posix/cpu,posix/mem > 2014-10-27 16:53:05,854:1(0x7fd859eee700):ZOO_INFO@log_env@712: Client > environment:zookeeper.version=zookeeper C client 3.4.5 > 2014-10-27 16:53:05,854:1(0x7fd859eee700):ZOO_INFO@log_env@716: Client > environment:host.name <http://host.name/>=slave-hostname > 2014-10-27 16:53:05,854:1(0x7fd859eee700):ZOO_INFO@log_env@723: Client > environment:os.name <http://os.name/>=Linux > 2014-10-27 16:53:05,854:1(0x7fd859eee700):ZOO_INFO@log_env@724: Client > environment:os.arch=3.16.2+ > 2014-10-27 16:53:05,854:1(0x7fd859eee700):ZOO_INFO@log_env@725: Client > environment:os.version=#2 SMP Wed Oct 1 20:08:48 UTC 2014 > 2014-10-27 16:53:05,854:1(0x7fd859eee700):ZOO_INFO@log_env@733: Client > environment:user.name <http://user.name/>=(null) > I1027 16:53:05.854481 1 main.cpp:149] Starting Mesos slave > I1027 16:53:05.854866 11 slave.cpp:167] Slave started on 1)@1.2.3.4:5051 > <http://1.2.3.4:5051/> > I1027 16:53:05.855069 11 slave.cpp:278] Slave resources: cpus(*):8; > mem(*):31188; disk(*):447410; ports(*):[31000-32000] > I1027 16:53:05.855099 11 slave.cpp:306] Slave hostname: slave-hostname > I1027 16:53:05.855110 11 slave.cpp:307] Slave checkpoint: false > I1027 16:53:05.856284 6 state.cpp:33] Recovering state from > '/tmp/mesos/meta' > I1027 16:53:05.856372 8 status_update_manager.cpp:193] Recovering status > update manager > I1027 16:53:05.856470 6 containerizer.cpp:252] Recovering containerizer > I1027 16:53:05.856484 9 docker.cpp:577] Recovering Docker containers > 2014-10-27 16:53:05,856:1(0x7fd859eee700):ZOO_INFO@log_env@741: Client > environment:user.home=/root > 2014-10-27 16:53:05,856:1(0x7fd859eee700):ZOO_INFO@log_env@753: Client > environment:user.dir=/tmp > 2014-10-27 16:53:05,856:1(0x7fd859eee700):ZOO_INFO@zookeeper_init@786: > Initiating client connection, host=MASTERS:2181 sessionTimeout=10000 > watcher=0x7fd85de56a30 sessionId=0 sessionPasswd=<null> > context=0x7fd828000f80 flags=0 > 2014-10-27 16:53:05,862:1(0x7fd856a8c700):ZOO_INFO@check_events@1703: > initiated connection to server [MASTER_IP:2181] > 2014-10-27 16:53:05,865:1(0x7fd856a8c700):ZOO_INFO@check_events@1750: session > establishment complete on server [REPLICA_IP:2181], > sessionId=0x1494d70090e0025, negotiated timeout=10000 > I1027 16:53:05.865543 7 group.cpp:313] Group process > (group(1)@1.2.3.4:5051 <http://1.2.3.4:5051/>) connected to ZooKeeper > I1027 16:53:05.865581 7 group.cpp:787] Syncing group operations: queue > size (joins, cancels, datas) = (0, 0, 0) > I1027 16:53:05.865608 7 group.cpp:385] Trying to create path '/mesos' in > ZooKeeper > I1027 16:53:05.866819 6 detector.cpp:138] Detected a new leader: (id='7') > I1027 16:53:05.866902 8 group.cpp:658] Trying to get > '/mesos/info_0000000007' in ZooKeeper > I1027 16:53:05.867444 7 detector.cpp:426] A new leading master > (UPID=master@MASTER_IP:5050) is detected > I1027 16:53:07.855305 13 slave.cpp:3198] Finished recovery > I1027 16:53:07.855576 12 slave.cpp:589] New master detected at > master@MASTER_IP:5050 > I1027 16:53:07.855669 12 slave.cpp:625] No credentials provided. > Attempting to register without authentication > I1027 16:53:07.855680 7 status_update_manager.cpp:167] New master > detected at master@MASTER_IP:5050 > I1027 16:53:07.855695 12 slave.cpp:636] Detecting new master > > The process exits right after the last line. > > If I don’t start the Docker containerizer everything works fine: > > $ docker run -ti --rm --name=mesos-slave --net=host -e > MESOS_ISOLATOR=cgroups/cpu,cgroups/mem -e MESOS_CONTAINERIZERS=mesos -e > MESOS_HOSTNAME=slave-host -e MESOS_IP=1.2.3.4 -v > /run/docker.sock:/run/docker.sock -v /sys:/sys -e MESOS_LOG_DIR=/var/log -e > MESOS_MASTER=zk://ZOOKEEPER_IPS:2181/mesos <> -p 5051:5051 -e > MESOS_CHECKPOINT=false mesos-slave > I1027 17:03:28.288579 1 logging.cpp:142] INFO level logging started! > I1027 17:03:28.288823 1 main.cpp:126] Build: 2014-09-23 05:36:09 by root > I1027 17:03:28.288838 1 main.cpp:128] Version: 0.20.1 > I1027 17:03:28.288849 1 main.cpp:131] Git tag: 0.20.1 > I1027 17:03:28.288857 1 main.cpp:135] Git SHA: > fe0a39112f3304283f970f1b08b322b1e970829d > I1027 17:03:28.290194 1 containerizer.cpp:89] Using isolation: > posix/cpu,posix/mem > I1027 17:03:28.290340 1 main.cpp:149] Starting Mesos slave > 2014-10-27 17:03:28,290:1(0x7f89ef493700):ZOO_INFO@log_env@712: Client > environment:zookeeper.version=zookeeper C client 3.4.5 > 2014-10-27 17:03:28,290:1(0x7f89ef493700):ZOO_INFO@log_env@716: Client > environment:host.name <http://host.name/>=slave-hostname > 2014-10-27 17:03:28,290:1(0x7f89ef493700):ZOO_INFO@log_env@723: Client > environment:os.name <http://os.name/>=Linux > 2014-10-27 17:03:28,290:1(0x7f89ef493700):ZOO_INFO@log_env@724: Client > environment:os.arch=3.16.2+ > 2014-10-27 17:03:28,290:1(0x7f89ef493700):ZOO_INFO@log_env@725: Client > environment:os.version=#2 SMP Wed Oct 1 20:08:48 UTC 2014 > 2014-10-27 17:03:28,290:1(0x7f89ef493700):ZOO_INFO@log_env@733: Client > environment:user.name <http://user.name/>=(null) > I1027 17:03:28.290735 8 slave.cpp:167] Slave started on 1)@1.2.3.4:5051 > <http://1.2.3.4:5051/> > I1027 17:03:28.290910 8 slave.cpp:278] Slave resources: cpus(*):8; > mem(*):31188; disk(*):447410; ports(*):[31000-32000] > I1027 17:03:28.290946 8 slave.cpp:306] Slave hostname: slave-hostname > I1027 17:03:28.290964 8 slave.cpp:307] Slave checkpoint: false > I1027 17:03:28.292132 13 state.cpp:33] Recovering state from > '/tmp/mesos/meta' > I1027 17:03:28.292201 12 status_update_manager.cpp:193] Recovering status > update manager > I1027 17:03:28.292268 12 containerizer.cpp:252] Recovering containerizer > I1027 17:03:28.292431 8 slave.cpp:3198] Finished recovery > 2014-10-27 17:03:28,292:1(0x7f89ef493700):ZOO_INFO@log_env@741: Client > environment:user.home=/root > 2014-10-27 17:03:28,292:1(0x7f89ef493700):ZOO_INFO@log_env@753: Client > environment:user.dir=/tmp > 2014-10-27 17:03:28,292:1(0x7f89ef493700):ZOO_INFO@zookeeper_init@786: > Initiating client connection, host=ZOOKEEPER_IPS:2181 sessionTimeout=10000 > watcher=0x7f89f33fba30 sessionId=0 sessionPasswd=<null> > context=0x7f89d0002580 flags=0 > 2014-10-27 17:03:28,296:1(0x7f89e7fff700):ZOO_INFO@check_events@1703: > initiated connection to server [REPLICA_IP:2181] > 2014-10-27 17:03:28,297:1(0x7f89e7fff700):ZOO_INFO@check_events@1750: session > establishment complete on server [REPLICA_IP:2181], > sessionId=0x3494d6e6dc00013, negotiated timeout=10000 > I1027 17:03:28.298100 7 group.cpp:313] Group process > (group(1)@1.2.3.4:5051 <http://1.2.3.4:5051/>) connected to ZooKeeper > I1027 17:03:28.298125 7 group.cpp:787] Syncing group operations: queue > size (joins, cancels, datas) = (0, 0, 0) > I1027 17:03:28.298138 7 group.cpp:385] Trying to create path '/mesos' in > ZooKeeper > I1027 17:03:28.299088 12 detector.cpp:138] Detected a new leader: (id='7') > I1027 17:03:28.299154 9 group.cpp:658] Trying to get > '/mesos/info_0000000007' in ZooKeeper > I1027 17:03:28.299595 10 detector.cpp:426] A new leading master > (UPID=master@MASTER_IP:5050) is detected > I1027 17:03:28.299659 8 slave.cpp:589] New master detected at > master@MASTER_IP:5050 > I1027 17:03:28.299717 8 slave.cpp:625] No credentials provided. > Attempting to register without authentication > I1027 17:03:28.299723 9 status_update_manager.cpp:167] New master > detected at master@MASTER_IP:5050 > I1027 17:03:28.299741 8 slave.cpp:636] Detecting new master > I1027 17:03:29.265413 12 slave.cpp:754] Registered with master > master@MASTER_IP:5050; given slave ID 20141027-040801-148971440-5050-1-5 > > And therefore I can see this slave on the web interface of the master. > > This is the Dockerfile of the container that I run: > > $ cat Dockerfile > FROM redjack/mesos-slave > > RUN apt-get install -y docker.io <http://docker.io/> > > In installed docker in the host to prevent the following error: > > Failed to create a containerizer: Could not create DockerContainerizer: > Failed to execute 'docker version': exited with status exited with status 127 > > I know that somebody managed to make it run in v0.19, but with v0.20 things > have changed a lot, as far as I have understood. Any thoughts? > > Thanks >
