2014년 12월 31일 수요일

[scm-user] Same host identification number is assigned to 2 hosts

I have been trying to install CM 5.3 on a 3-node cluster (the hosts are Virtual Box images).   

/etc/hosts

10.0.1.3  namenode3.cdh.com namenode3
10.0.1.4  secondarynamenode4.cdh.com secondarynamenode4
10.0.1.5  datanode5.cdh.com datanode5


 When I got to the point of adding new hosts to clusters, I saw only 2 out of the 3 hosts listed in the cluster.  Either secondarynamenode4 or datanode5 is shown in the list.  Upon looking at the /var/log/cloudera-scm-agent/, I find both of those hosts assigned the same host identification number.

/var/log/cloudera-scm-agent/cloudera-scm-agent.log on datanode5

30/Dec/2014 04:03:26 +0000] 2979 MainThread agent        INFO     SCM Agent Version: 5.3.0
[30/Dec/2014 04:03:26 +0000] 2979 MainThread agent        INFO     Agent Protocol Version: 4
[30/Dec/2014 04:03:26 +0000] 2979 MainThread agent        INFO     Using Host ID: f8091847-d4ed-48f9-92ab-c6504a9289b5  <===== 
[30/Dec/2014 04:03:26 +0000] 2979 MainThread agent        INFO     Using directory: /var/run/cloudera-scm-agent
[30/Dec/2014 04:03:26 +0000] 2979 MainThread agent        INFO     Using supervisor binary path: /usr/lib64/cmf/agent/src/cmf/../../build/env/bin/supervisord


/var/log/cloudera-scm-agent/cloudera-scm-agent.log on secondarynamenode4

[30/Dec/2014 01:36:45 +0000] 2913 MainThread agent        INFO     SCM Agent Version: 5.3.0
[30/Dec/2014 01:36:45 +0000] 2913 MainThread agent        INFO     Agent Protocol Version: 4
[30/Dec/2014 01:36:45 +0000] 2913 MainThread agent        INFO     Using Host ID: f8091847-d4ed-48f9-92ab-c6504a9289b5 <===== 
[30/Dec/2014 01:36:45 +0000] 2913 MainThread agent        INFO     Using directory: /var/run/cloudera-scm-agent
[30/Dec/2014 01:36:45 +0000] 2913 MainThread agent        INFO     Using supervisor binary path: /usr/lib64/cmf/agent/src/cmf/../../build/env/bin/supervisord


The question I have is how was the same host id generated or assigned to 2 different hosts added to a cluster.  Is there anything on the host I can configure to fix this problem?  Should I file a bug?





Delete the hosts and start over in CM, since you are in the very beginning stage.

validate the hosts file on all the 3 servers to reflect the same ip address respectively to the hosts. If the ip address  are not correct you can get j to this situations because different CM agents report back to CM server incorrectly.



Thanks for the reply and happy new year!  

What prompts to the original question was that the same hosts and host files worked in CM 5.2 but failed in CM 5.3.  And fyi, CM did reported 2 different IP addresses but yet they have the same UUID, the host identification number.  So the original question of how was the same host id generated or assigned to 2 different hosts added to a cluster is still unanswered.



Good question I am not sure how it happened  in your case when I upgraded my 10 node test cluster everyeverything was smooth and mine has security enabled and integrated to AD sentry enabled.

I hope some CM knowledge Folks can answer it much better than me try posting the same question in that forum too.


댓글 없음:

댓글 쓰기