Re: just added a new real server, getting a really large number of InActConn

[Graeme Fowler <graeme@xxxxxxxxxxx>]

Hi

On Tue 07 Feb 2006 16:02:24 GMT , Randy Paries <rtparies@xxxxxxxxx> wrote:
> I just added a new Real Server to my cluster
>
> local.lovejoy is the one i just added.
<snip>
> ipvsadm
> IP Virtual Server version 1.0.10 (size=65536)
> Prot LocalAddress:Port Scheduler Flags
>  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
> TCP  www.unitnet.com:http wlc persistent 1800 mask 255.255.255.0
>  -> local.lovejoy:http           Route   1      113        5568
>  -> local.krusty:http            Route   1      97         223
>  -> local.flanders:http          Route   1      91         158
> TCP  www.unitnet.com:https wlc persistent 1800 mask 255.255.255.0
>  -> local.flanders:https         Route   1      0          12
The newer machine is a newer OS, running a newer version of Apache and 
probably newer hardware (OK, those last two are assumptions) - I bet 
it's responding more quickly.
The InActConn counter diaplays thso connection in TIME_WAIT or related 
states, after a FIN packet has arrived to end the connection. If you 
run:
ipvsadm -Ln --persistent-conn --sort --rate

and ipvsadm -Ln --persistent-conn --sort --stats

You will probably see that lovejoy is handling rather more traffic than 
krusty and flanders.
Graeme