WebSocket queue is full, dicarding 'metrics' message

asked 2018-11-09 09:47:00 -0500

supahcraig gravatar image

Can anyone help me understand what this message is telling me, and what the cause/resolution is? The pipeline consists of a Kafka origin writing to the HDFS FS destination. The pipeline has been running for 9 days and appears to still be successfully running based on the fact that I still see data written to the destination.

So why is the websocket queue full, and what is this metrics message that is being discarded? And how do I rectify it?

edit retag flag offensive close merge delete


Do you have pipeline metric rules defined? As per the docs here: https://streamsets.com/documentation/datacollector/latest/help/datacollector/UserGuide/Alerts/RulesAlerts_title.html If so, what values do you have set for Metric Element when you expand them?

jeff gravatar imagejeff ( 2018-11-09 10:04:51 -0500 )edit

I'm not sure I completely know how to answer your question, but I didn't set up any metrics. There are 5 there, which I assume are default metric rules? Two are related to errors, but I don't have any errors. One is related to an idle pipeline, one for memory, and one for slow batch.

supahcraig gravatar imagesupahcraig ( 2018-11-09 11:53:53 -0500 )edit