Forum Discussion
NimbostratusHTTP Round Robin not working from browser
I have a very simple LAB setup with F5 LTM 11.3 Trial and two IIS web servers. All virtualized with VMWare Workstation 10. One virtual server (HTTP) with two nodes (HTTP) and round robin load balancing. Each IIS server contains a simple /test.html page indicating which node is answering HTTP request. (Node 01 or Node 02). No iRules, No Group Priority. No monitors.
Test 1: When refreshing Google Chrome against Virtual Server (HTTP) test page does not change. Even with "Empty buffer, Hard reload" option i Chrome. But connections are distributed even between pool members when viewing pool statistics. Same thing with Internet explorer. I dont understand why this is happening!
Test 2: Testing with a telnet against Virtual server on port 80 and a "GET /test.html HTTP/1.1" works perfectly. Node 1 answears HTTP request first, Node 2 answears second, Node 1 again third etc. Pool statistic shows a even distributed connections betweens pool memebers. This is what I expect should happen when using Chrome or Internet Explorer.
I am new to F5 and proberly missing something here, but what?
4 Replies
Hi!
The load balancing is done per connection, not per request. Browsers uses keep-alives and reuses connections while telnet closes the connection after getting the response.
Hope that answers your question?
/Patrik
JT010_161778Yes it does, thanks.
Followup question: Ia there any way to force browsers to open a new connection similar to telnet?
If you disable HTTP keepalive it might work. May I ask why you would want this?
/Patrik
- JT010_161778
No good reason really, just for testing/Proof of concept. Telnet ofcourse does work as expected but it would be simpler using a browser :-)
