I am using nagios to monitor Hadoop cluster.
Would like to hear input from you guys.
1. Would that be any difference between monitoring
TCP port 9000 and curl port 50070 and grep for "namenode"
2. Job tracker I will monitor tcp port 9001 any drawbacks ?
3. Secondarynamenode what would be the good way to monitor it ?
- process if it is up and running
- if fsimage is outdate
----> input are more than welcome..
- tcp check port ?
Grokbase › Groups › Hadoop › common-user › September 2011