LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: problem with too many inactive connections

To: Julian Anastasov <uli@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: problem with too many inactive connections
Cc: lvs-users@xxxxxxxxxxxxxxxxxxxxxx, Joseph Mack <mack@xxxxxxxxxxx>
From: Wayne <wayne@xxxxxxxxxxxxxxx>
Date: Wed, 17 May 2000 11:35:24 -0700
I am upgrading to 2.2.15 OS, but the Intel network card
driver somehow not working properly -- ifconfig hangs
during bring the interface up.  I remember seeing one
report of similar problem on this list.  Did anyone else
has problem running Intel pro 10/100 card at 2.2.15
OS level?  It was working fine at 2.2.13.

At 02:27 PM 5/16/00 -0700, Wayne wrote:
>As one of the person in the list responded yesterday, I am running
>2.2.13 kernel and 0.9.7 LVS.  The other person pointed out that
>the problem was in the 2.2.13 kernel.  So I am upgrading the
>kernel to 2.2.15 now to see if the problem will go away :)
>
>Thanks for the help!
>
>At 07:02 AM 5/16/00 +0300, Julian Anastasov wrote:
>
> >         Hello,
> >
> >On Mon, 15 May 2000, Wayne wrote:
> >
> > > I just noticed that when one servers InactivConn number is
> > > about 4000 reported by ipp -l in a two server farm, that
> > > server will not get any further traffic, even after that server
> > > becoming okay.
> > > 
> > > Does anyone see this kind problem before?  I am running
> > > a older version of LVS kernel, but I thought that could be
> > > the same for new ones, if no one reported and fixed it.
> > > 
> > > Thanks!
> >  
> >         You have to give more details. I don't remember for
> >such problem. But if you have some problems with the real
> >servers in VS/DR or VS/TUN mode, sometimes the number of
> >active/inactive connections are wrong. This can happen when
> >some of the real servers are deleted and the traffic is
> >temporary stopped. In these modes the director handles only
> >the input traffic and these numbers represent the connection
> >state in the time of deleting the real servers.
> >
> >         But you have to give some information or to
> >investigate the problem, in what conditions it is raised.
> >
> >
> >Regards
> >
> >--
> >Julian Anastasov <uli@xxxxxxxxxxxxxxxxxxxxxx>
>
>



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