Basic round robin question

[bjelf]

Hi there, I have a cisco css-11506 running 7.50.1.03. One of our http content rules has 3 services using round robin and no sticky. Due to performance issues recently, 2 of the 3 servers had their tcp stacks tweaked to improve server to database communication. One was left unchanged for comparison. The 2 servers changed are running faster than the third server. Now when I look at hits/conns in the content rule, the 3rd unchanged server has far more of each:

Name: Hits: Wgt: State: Ld: KAlive: Conn: DNS:

----- ----- ----- ------ --- ------- ----- ----

srv1-80 1,067,413 S-1 Alive 28 TCP-80 919 0

srv2-80 1,020,149 S-1 Alive 34 TCP-80 966 0

srv3-80 1,572,125 S-1 Alive 41 TCP-80 1151 0

I understand why there would be more active connections, the server is processing things slower than the other two. But why would the hits appear much higher on the slow server? Seems like in simple round robin this should be pretty much equal. Could it be http1.1 sticky and retransmits on the slow server?

Thanks

-Dan

[Gilles Dufour]

Dan,

there is no explanation for this.

Was there any service transition ?

could you reset the counter [with the 'zero' command] and does it still show the same behavior ?

can we get a 'show service summary' and the complete config ?

Thanks,

Gilles.