LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

[lvs-users] Fwd: InActConn dilemma

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: [lvs-users] Fwd: InActConn dilemma
From: Andre Magri <andremag@xxxxxxxxx>
Date: Sat, 13 Feb 2010 20:48:44 +0100
Hi again,
Is there a logical reason as to why one of the servers is having so many
connections being turned into inactive (FIN_WAIT) so quickly?

I have read that ipvsadm should keep connections established for a while but
I simply cannot understand why the load balancer is behaving like this.  Any
clues will be appreciated.

Thanks again.

---------- Forwarded message ----------
From: Andre Magri <andremag@xxxxxxxxx>
Date: 12 February 2010 17:10
Subject: Re: [lvs-users] InActConn dilemma
To: "LinuxVirtualServer.org users mailing list." <
lvs-users@xxxxxxxxxxxxxxxxxxxxxx>


ipvsadm -Ln --stats

IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port               Conns   InPkts  OutPkts  InBytes
OutBytes
  -> RemoteAddress:Port
FWM  80                            2053990 19742182        0    4960M
0
  -> www2:80                   59309  3268078        0  896604K        0
  -> www1:80                 1994681 16474104        0    4064M        0

Except for the NIC team the webservers are configured exactly the same.

Every 1.0s: ipvsadm -Ln
--rate
Fri Feb 12 17:08:59 2010


IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port                 CPS    InPPS   OutPPS    InBPS
OutBPS
  -> RemoteAddress:Port
FWM  80                                134     1280        0   369766
0
  -> www2:80                       4      214        0    45576        0
  -> www1:80                     130     1066        0   324189        0

Am I going to have a problem when heavy traffic hits?

Thanks!




On 12 February 2010 16:33, Graeme Fowler <graeme@xxxxxxxxxxx> wrote:

> On Fri, 2010-02-12 at 16:25 +0100, Andre Magri wrote:
> > IP Virtual Server version 1.2.1 (size=4096)
> > Prot LocalAddress:Port Scheduler Flags
> >   -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
> > FWM  80 wlc persistent 1200
> >   -> www2.mysite.com:http        Route   1      359        146
> >   -> www1.mysite.com:http        Route   1      127        *12176*
> >
> > The high rate of InActConn for www1 is worrying me.
>
> That indicates one of several possible things:
>
> 1. www1 is handling more connections than www2, or
> 2. www1 is not closing connections properly
>
> Are the webservers configured in the same way?
>
> What do you see if you do:
>
> ipvsadm -Ln --stats
>
> Also, if you do this:
>
> watch -n1 -- ipvsadm -Ln --rate
>
> [use ctrl-c to exit]
>
> what do you see? I'd surmise that www1 is handling a lot more traffic
> than www2.
>
> Graeme
>
>
> _______________________________________________
> Please read the documentation before posting - it's available at:
> http://www.linuxvirtualserver.org/
>
> LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
> Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
> or go to http://lists.graemef.net/mailman/listinfo/lvs-users
>
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/

LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
or go to http://lists.graemef.net/mailman/listinfo/lvs-users

<Prev in Thread] Current Thread [Next in Thread>