Pankaj260100 commented on PR #17100: URL: https://github.com/apache/druid/pull/17100#issuecomment-2454476043
By adding a log line whenever threads block waiting for the channel, I observed that the number of waiting threads increases during periods of high latency. After adding this metric to track the channel wait time, I found it typically stays within 2-3ms. However, during high latency periods, this wait time increases to as much as 9 seconds. The number of open connections at the data nodes remains stable, likely due to the default 4-minute unusedConnectionTimeout. One possible solution is increasing druid.broker.http.numConnections. Alternatively, there might be an issue at the data nodes delaying query processing and causing requests to wait for connections at the broker. cc: @kfaraz @LakshSingla @abhishekagarwal87 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
