In our splunk deployment, we have about 100 universal forwarders installed on PCs and forward data to a splunk server (port: 998) that will further forward the data to another splunk server. The intermediate splunk server also serves as a deployment server (port: 8089) to deploy apps to the PCs. We tested this on about 20 machines initially and it worked fine. But after we expanded the deployment to the rest of the machines, we noticed the intermediate splunk server is rejecting the connections from the clients, citing SSL error.
ERROR TcpInputFd - SSL Error for fd from HOST:x.x.x.x, IP:x.x.x.x, PORT:2933
We looked at the splunk log at one of the PCs. The log clearly said that it attempts to open a non-SSL connection "INFO TcpOutputProc - Initializing connection for non-ssl forwaring to x.x.x.x". We are puzzled that why the error is related to SSL at the receiver end.
The first step would be to look over your inputs.conf (on the UF) and outputs.conf (on the IF)configurations for this setup. If you could post the relative details here, we could help you troubleshoot this further.
We figured out that the SSL error is related to deployment server. Once we disable the connection to the deployment server at the client end, we no longer see this error at the server end.
However, not all the clients get connected the splunk server. We found this error message at the client end.
"Connect to x.x.x.x:y failed. No connection could be made because the target machine actively refused it."
It seems to suggest that the intermediate splunk server rejects the connection request from client. However, we cannot locate any information that tells me the cause of it. Any advice?