We have the following network architecture:
Public facing F5 -> [HTTPS] -> Apache HTTPD Server with mod_ssl -> [Connects using reverse proxy HTTPS with mod_proxy + SSLProxyEngine] -> Internal facing F5 -> Various WebLogic Servers and IIS servers
Note: Apache in the middle has SSLEngine off. It is simply an SSL client that uses SSLProxyEngine to make an HTTPS connection to the internal facing F5 using mod_proxy/mod_ssl.
There are two different virtual IPs that we connect through.
Traffic to the IIS servers, no problem. Traffic to the Weblogic servers, we get a 502 error for only ONE of the virtual IPs.
We ran tcpdump then reproduced the issue on the apache server. That and the error logs both show that the Internal F5 sent an unexpected TCP reset packet.
Has anyone ever experienced this happening with just one VIP? I am pretty convinced this is all based on F5 behaving badly. Would you agree?
Since the only segment with an issue is between f5 -> weblogic servers (one of the VIP), please provide more details about it. Is the f5 pool containing weblogic servers showing the servers in a healthy state ? Does most of the communication work but only certain fail after say a certain timeout ? Can you access the weblogic servers directly without any problems ?