NimbostratusI have been working on an issue where pool members (web servers on port 80) is getting GET and HEAD messages from the self-IPs from a cluster and from floating IP. Even after I changed the Health Monitor from default to a custom one, the web servers are still receiving almost 500 requests in 10 seconds. Any suggestions?
Nimbostratus500 per 10 seconds is not realistic. That would be 0.02 second interval in between the HTTP health check requests to a single pool member. Interval this short would cause each pool member to flip available/unavailable even in ideal LAN conditions.
What you're seeing is not traffic that originates from F5 BigIP. It's traffic that F5 sends to pool members after normal balancing/routing procedure. That traffic originates from elsewhere. Is one or more of your Virtual Servers configured to use SNAT automap or SNAT Pool feature? That would explain a packet capture and/or server logs where BigIP's SelfIP is seen as the client.
AltostratusHi,
You can increase monitor polling time for your pool. which will decrease the monitor polling requests from F5 to server. If you want to see real IP then you can disable SNAT but you need to see other options like routes etc.. before disabling SNAT option on virtual server.
Hope this helps.