CODEWITHSUNDEEP
dockerapache-flinkmesosmarathon
M_Gh
M_Gh

Reputation: 1142

Flink cannot be run in Marathon

I have three physical nodes with docker installed on them. I have one docker container with Mesos, Marathon, Hadoop and Flink. I configured Master node and Slave nodes for Mesos,Zookeeper and Marathon. I do these works step by step. First, In Master node, I enter to docker container with this command:

docker run -v /home/user/.ssh:/root/.ssh --privileged -p 5050:5050 -p 5051:5051 -p 5052:5052 -p 2181:2181 -p 8082:8081 -p 6123:6123 -p 8080:8080 -p 50090:50090 -p 50070:50070 -p 9000:9000 -p 2888:2888 -p 3888:3888 -p 4041:4040 -p 7077:7077 -p 52222:22 -e WEAVE_CIDR=10.32.0.2/12 -e MESOS_EXECUTOR_REGISTRATION_TIMEOUT=5mins -e LIBPROCESS_IP=10.32.0.2 -e MESOS_RESOURCES=ports*:[11000-11999] -ti hadoop_marathon_mesos_flink_2 /bin/bash

Then run Mesos and Zookeeper :

/home/zookeeper-3.4.14/bin/zkServer.sh restart
/home/mesos-1.7.2/build/bin/mesos-master.sh --ip=10.32.0.1 --hostname=10.32.0.1 --roles=marathon,flink --quorum=1 --work_dir=/var/run/mesos --log_dir=/var/log/mesos

After that run Marathon in the same container:

 /home/marathon-1.7.189-48bfd6000/bin/marathon --master 10.32.0.1:5050 --zk zk://10.32.0.1:2181/marathon --hostname 10.32.0.1 --webui_url 10.32.0.1:8080 --logging_level debug

And finally, I run hadoop:

  /opt/hadoop/sbin/start-dfs.sh

Marathon,Mesos and Hadoop are run without any problems. The most important part of my work is running Flink in Marathon. I configured Flink in docker container like this:

  env.java.home: /opt/java
  jobmanager.rpc.address: 10.32.0.1
  high-availability: zookeeper
  high-availability.storageDir: hdfs:///flink/ha/
  high-availability.zookeeper.quorum: 10.32.0.1:2181,10.32.0.2:2181
  recovery.zookeeper.path.mesos-workers: /mesos-workers

In Marathon UI, I create Application and put this JSON file on it, but it is failed.

  {
  "id": "flink",
  "cmd": "/home/flink-1.7.0/bin/mesos-appmaster.sh 
   -Dmesos.master=10.32.0.1:5050,10.32.0.2:5050 
   -Dmesos.initial-tasks=1",
  "cpus": 1.0,
  "mem": 1024
   }

Flink application is failed in Mesos UI. It shows this error:

I0428 06:01:39.586699 6155 exec.cpp:162] Version: 1.7.2

I0428 06:01:39.596458 6154 exec.cpp:236] Executor registered on agent 984595ae-e811-48fb-a9f5-ca6128e1cc1a-S0

I0428 06:01:39.598870 6157 executor.cpp:188] Received SUBSCRIBED event

I0428 06:01:39.599761 6157 executor.cpp:192] Subscribed executor on 10.32.0.3

I0428 06:01:39.599963 6157 executor.cpp:188] Received LAUNCH event

I0428 06:01:39.601236 6157 executor.cpp:697] Starting task flink.16a7cc18-697b-11e9-928f-ce235caa831e

I0428 06:01:39.613719 6157 executor.cpp:712] Forked command at 6163

I0428 06:01:39.787395 6157 executor.cpp:1013] Command exited with status 1 (pid: 6163)

I0428 06:01:40.791885 6162 process.cpp:927] Stopped the socket accept loop

The strange thing is that in STDout, I see this text; even though I set JAVA_HOME in /etc/environment and flink-conf.yam.

Please specify JAVA_HOME. Either in Flink config ./conf/flink-conf.yaml or as system-wide JAVA_HOME.

Would you please tell me what I should do for that problem?

Many Thanks.

Upvotes: 0

Views: 894

Answers (1)

r.a.shehni
r.a.shehni

Reputation: 378

You can check your Flink log in Slave node. Also, it is better to change your JSON file like this. It helps you to follow your application.

 {
  "id": "flink",
  "cmd": "/home/flink-1.7.0/bin/mesos-appmaster.sh -Djobmanager.heap.mb=1024 
  -Djobmanager.rpc.port=6123 -Drest.port=8081 
  -Dmesos.resourcemanager.tasks.mem=1024 -Dtaskmanager.heap.mb=1024 
  -Dtaskmanager.numberOfTaskSlots=2 -Dparallelism.default=2 
  -Dmesos.resourcemanager.tasks.cpus=1",
   "cpus": 1.0,
   "mem": 1024,
   "fetch": [
     {
      "uri": "/home/flink-1.7.0/bin/mesos-appmaster.sh",
      "executable": true
     }
      ]
    }

Also, JAVA_HOME to Flink_conf.yaml in every nodes, Master and Slaves.

 env.java.home: /opt/java

With adding JAVA_HOME, you do not see the error in STDOUT.

I think it is useful.

Upvotes: 1

Related Questions