We are running a standalone deployment server version 4.3.5.
Using the search "index=_* hostname=SplunkInfrastructureHost* fwdType=uf OR fwdType=lwf |stats count by fwdType, hostname "
our deployment server is being reported as being a "lwf".
Don't remember seeing this when running an earlier version of Splunk. How does Splunk determine the fwdType?
Thanks.
The 'fwdType' field is read from metrics.log events for group=tcpin_connections, recording activity and information on a per-forwarder basis. Here's an example:
02-20-2013 21:20:02.316 -0500 INFO Metrics - group=tcpin_connections, 20.31.175.123:2783:9997, connectionType=cooked, sourcePort=2783, sourceHost=thing.example.net, sourceIp=20.31.175.123, destPort=9997, _tcp_Bps=96.75, _tcp_KBps=0.09, _tcp_avg_thruput=0.29, kb=2.94, _tcp_Kprocessed=12246.86, _tcp_eps=0.13, build=140868, version=5.0, os=Windows, arch=Intel, hostname=thing, guid=AB557B33-AF37-4D9E-BE4C-D6F3F859E1C9, fwdType=uf, ssl=false, lastIndexer=20.31.175.89:9997, ack=false
It does seem somewhat surprising that your Deployment Server is registered as 'lwf' given that if it is indeed sending it's _internal index to your indexers, it is doing this as a regular forwarder.
You may want to look for your DS' IP and/or hostname in metrics.log and see what value of 'lwf' is registered there for the corresponding group=tcpin_connections events.
Seems the problem is that the IP address for the DS is being NAT'ed behind an F5 so that is causing the problem. Must be something our network folks changed not too long ago.
Thanks for the response.