[
https://issues.apache.org/jira/browse/FLINK-9649?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16528023#comment-16528023
]
Leonid Ishimnikov commented on FLINK-9649:
------------------------------------------
[~gjy] Thank you for the explanation.
Perhaps Mesos-related documentation is a bit ambiguous after the introduction
of FLIP-6:
{quote}mesos.initial-tasks: determines the initial workers to bring up when the
master starts
{quote}
For a newcomer or someone upgrading from pre-1.5, it's not very clear which
master is being referred to. If it's the Mesos Application Master that is
started with {{mesos-appmaster.sh}} script, which would be my first assumption,
then this is incorrect. It sounds more like it would refer to the Job Master
(Mesos equivalent of YARN's Application Master). Would it make sense to
specify in the documentation that this refers to Job Master?{{}}{{}}
> TaskManagers are not scheduled on Mesos
> ---------------------------------------
>
> Key: FLINK-9649
> URL: https://issues.apache.org/jira/browse/FLINK-9649
> Project: Flink
> Issue Type: Bug
> Components: Mesos
> Affects Versions: 1.5.0, 1.6.0
> Reporter: Leonid Ishimnikov
> Priority: Major
>
> Flink correctly registers as a framework, but does not schedule task managers.
> Command:
> {noformat}
> ./bin/mesos-appmaster.sh -Dmesos.master="zk://192.168.0.101:2181/mesos"
> -Djobmanager.heap.mb=1024 -Djobmanager.rpc.address=$(hostname -i)
> -Djobmanager.rpc.port=6123 -Djobmanager.web.address=$(hostname -i)
> -Djobmanager.web.port=8080 -Dmesos.initial-tasks=2
> -Dmesos.resourcemanager.tasks.mem=4096 -Dtaskmanager.heap.mb=3500
> -Dtaskmanager.numberOfTaskSlots=2 -Dparallelism.default=10
> -Dmesos.resourcemanager.tasks.cpus=1
> -Dmesos.resourcemanager.framework.principal=someuser
> -Dmesos.resourcemanager.framework.secret=somepassword
> -Dmesos.resourcemanager.framework.name="Flink-Test"{noformat}
> Log:
> {noformat}
> 2018-06-22 17:39:27,082 INFO
> org.apache.flink.runtime.entrypoint.ClusterEntrypoint - TaskManagers
> will be created with 2 task slots
> 2018-06-22 17:39:27,082 INFO
> org.apache.flink.runtime.entrypoint.ClusterEntrypoint - TaskManagers
> will be started with container size 4096 MB, JVM heap size 2765 MB, JVM
> direct memory limit 1331 MB, 1.0 cpus, 0 gpus
> ...
> 2018-06-22 17:39:27,304 INFO
> org.apache.flink.runtime.resourcemanager.slotmanager.SlotManager - Starting
> the SlotManager.
> 2018-06-22 17:39:27,305 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Registering as new framework.
> 2018-06-22 17:39:27,305 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> --------------------------------------------------------------------------------
> 2018-06-22 17:39:27,305 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Mesos Info:
> 2018-06-22 17:39:27,305 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Master URL: zk://192.168.0.101:2181/mesos
> 2018-06-22 17:39:27,305 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Framework Info:
> 2018-06-22 17:39:27,305 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> ID: (none)
> 2018-06-22 17:39:27,305 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Name: Flink-Test
> 2018-06-22 17:39:27,305 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Failover Timeout (secs): 10.0
> 2018-06-22 17:39:27,305 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Role: *
> 2018-06-22 17:39:27,306 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Capabilities: (none)
> 2018-06-22 17:39:27,306 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Principal: someuser
> 2018-06-22 17:39:27,306 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Host: 192.168.0.100
> 2018-06-22 17:39:27,306 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> Web UI: (none)
> 2018-06-22 17:39:27,306 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager -
> --------------------------------------------------------------------------------
> 2018-06-22 17:39:27,432 INFO
> org.apache.flink.mesos.scheduler.ConnectionMonitor - Connecting to
> Mesos...
> 2018-06-22 17:39:27,434 INFO
> org.apache.flink.mesos.runtime.clusterframework.MesosResourceManager - Mesos
> resource manager initialized.
> 2018-06-22 17:39:27,444 INFO
> org.apache.flink.runtime.dispatcher.StandaloneDispatcher - Dispatcher
> akka.tcp://[email protected]:6123/user/dispatcher was granted leadership
> with fencing token 00000000000000000000000000000000
> 2018-06-22 17:39:27,444 INFO
> org.apache.flink.runtime.dispatcher.StandaloneDispatcher - Recovering
> all persisted jobs.
> 2018-06-22 17:39:27,466 INFO
> org.apache.flink.mesos.scheduler.ConnectionMonitor - Connected to
> Mesos as framework ID 7295a8f7-c0a9-41d1-a737-ae71c57b72bf-1141.{noformat}
> There is nothing further in the log after that.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
