The nodemanager is going in a bad state and this is the message. If I try to restart YARN, the status is green for a while before it gets red again (I can run one or two pig scripts before they stop running). I have checked the logs and there is no error of any kind. Where can I start debugging this? Can Pig cause YARN to go into bad health?
I don't believe that a user job can cause YARN to fail. When the status is in that bad state, can you run the Host Inspector in CM? Also, I find it very surprising that there isn't any error in the logs. Did you check both the RM log and the NM log?
Yes I did check both the RM and NM logs. The issue turned out to be the pig version - things worked when I went from Pig 0.12 to trunk, the nodemanager continues to be in the bad state. But things are working.