Operator slowness & bad data counts

Hi Alpine,

Some of my users have been experiencing performance problems with some of their MR operators. I looked into the logs on the Hadoop cluster, and it seems that there is a bad data reporting timeout problem being reported in the logs for the job (I guess its waiting for this timeout to occur that is causing the slowdown?). Looking at the Chorus URL specified in the error message, it looks like its trying to use an external IP address for Chorus, rather than the needed internal IP address.

Is there a way to get Chorus to use the correct IP address?

TIA

(1) Answer

Login