Upgraded Deployment Server / License Server to Version 6. Runs OK for 15 minutes then maxes out CPU and starts to fail with
ERROR TcpChannel - Error trying to begin socket accept: An invalid argument was supplied.
showing in SplunkD log.
Are you using IPV6?
I had a similar issue with Splunk 6.0 build 182037. "ERROR TcpChannel - Error trying to begin socket accept: An invalid argument was supplied."
I pinged the Splunk Indexer and found that it was using IPV6 to resolve the hostname. I disabled IPV6, then added the hostname IPV4 address in the localhosts file. Restarted Splunk and it is working fine now.
I'm experiencing this problem too.. happening among all of my indexers and all of my heavy forwarders. Clients of each type are getting "WARN TcpOutputProc - Cooked connection to ip=x.x.x.x:9997 timed out". Problem continues even after upgrading from 6.0 and 6.0.3