index delays with universal forwarder's parsing and tcpout queue full

[cwl]

We are using 5.0.3 universal forwarder to send data to our 5.0.3 cluster environment.

We have enabled useACK because it is recommended as mentioned in splunk's documentation when using cluster but after enabling useACK we are seeing's our universal forwarder's parsing queue and tcpout queue quite full and data is getting indexed with delays.

We have already set thruput value in limits.conf to "maxKBps = 0" and we have checked with our network administrator that we don't have any network issues.

Our indexers seem to be working fine without any full queues. How can I fix this issue ?

[cwl · Accepted solution]

When you enable useACK, you have to increase your forwarder's maxQueueSize to 7MB in outputs.conf to ensure your forwarder does not get blocked while waiting for acknowledgment from an indexer.

[tcpout] 

maxQueueSize = 7MB 

Please refer below documentation for details and note that this is adjusted automatically in 5.0.4 and beyond.
link:Configure the wait queue size

[dfrankekcg]

Per http://docs.splunk.com/Documentation/Splunk/6.1.1/Forwarding/Protectagainstlossofin-flightdata, maxQueueSize=auto is recommended for forwarder version 5.04 and above.