Cluster mode connect error

asked 2018-03-31 23:17:52 -0500

davidha gravatar image

updated 2018-04-10 20:17:04 -0500

I am having a few pipelines scheduled to run every day in cluster batch mode. I have ran successfully for few days. But suddenly CONNECT_ERROR keep coming up and all the scheduled job I have just hanged there. What is the root cause of that? How do I avoid it or is there any work around? Found a post describing the bug but any ideas how the bug will be triggered?

edit retag flag offensive close merge delete

Comments

1

Are you referring specifically to this bug? If so, the workaround is described in the Jira (as well as release notes you linked): https://issues.streamsets.com/browse/SDC-2374 Can you confirm the exact SDC version you are using please?

jeff gravatar imagejeff ( 2018-04-03 15:09:00 -0500 )edit

I am using 2.7.2.0 version. We are using Streamsets. I saw the work around describe in post is to use force stop, the problem is we can't tell whether or not the last job loaded data. I experienced that even "Connect_error" appeared, the data have already loaded to the destination.

davidha gravatar imagedavidha ( 2018-04-03 21:13:26 -0500 )edit

The change status in file nor the force stop helps in the cluster mode, any ideas? I am currently restarting the whole SDC to get around this

davidha gravatar imagedavidha ( 2018-04-10 09:47:08 -0500 )edit