|
I guess it might be a version issue, 'ivsadm -v' yields:
ipvsadm v1.24 2003/06/07 (compiled with popt and IPVS v1.2.0)
On 8/1/07, Dan Baughman <dan.baughman@xxxxxxxxx> wrote:
>
> Thanks for the suggestions so far, here is what the dumps yielded.
>
> I'm using ethereal to look at a side-by-side comparison of a healthy
> conversation and one of the delayed conversations. The
> delayed conversation has two tcp packet types that the normal conversation
> doesn't:
> 1) A lot of duplicate acks from the client to the server
> 2) TCP Retransmissions from the server to client.
>
> It seems that there is a lot of traffic being duplicated and received by
> the real server twice. This isn't occuring when they access the realserver
> ip to view the site. The extra load time is occuring after the server
> receives such a duplicate ack, its a pause of about 3 seconds, then a tcp
> retransmission occurs.
>
> ~Dan
>
>
> On 7/31/07, Gerry Reno <greno@xxxxxxxxxxx> wrote:
> >
> > Gerry Reno wrote:
> > > Dan Baughman wrote:
> > >
> > >> I am trying to gather a sniffing session from both of my sides of the
> >
> > >> connection now. So far, they have both been comcast users.
> > >>
> > >> On 7/31/07, Joseph Mack NA3T < jmack@xxxxxxxx> wrote:
> > >>
> > >>
> > >>> On Tue, 31 Jul 2007, Dan Baughman wrote:
> > >>>
> > >>>
> > >>>
> > >>>> There are some users that when acessing my page through
> > >>>> the load balancer experience a 45 second load time, but
> > >>>> when going directly to either server takes only ten
> > >>>> seconds. I'm doing one nic, two network type load
> > >>>> balancing. (we have one loadbalancer and two real servers)
> > >>>>
> > >>>>
> > >>> this mysteriously happens sometimes. There's always some
> > >>> rational explanation after the fact, but it isn't obvious at
> > >>> the time. It happens because there are 3 nodes involved in
> > >>> making a 2 node connection. Things that are taken for
> > >>> granted in a 2 node connection don't always work in a 3 way
> > >>> connection.
> > >>>
> > >>> You may have to feret it out yourself. Possible things are
> > >>>
> > >>> o someone has to do DNS and it's failing
> > >>>
> >
> > Joe has a good point here. We do not have DNS but I am beginning to
> > wonder in our case if something isn't trying to do reverse lookup for
> > the IP and then timing out.
> >
> > >>> o you have identd on the realserver (turn it off)
> > >>>
> > >>> o there's some routing wierdness and icmp's etc are having
> > >>> to figure out the right path.
> > >>>
> > >>>
> > >>>
> > > I'm seeing this problem as well. It is mainly on first logins after
> > some
> > > period of time. It will take anywhere from 45-60 secs for the user to
> > > get logged in and then everything seems fine. I had some delays before
> > > this too but when I restarted apache that time everything was ok after
> >
> > > that. Now I am not so sure that it was apache because I also restarted
> > > keepalived.
> > >
> > >
> > > Gerry
> > >
> > >
> > > _______________________________________________
> > > 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
> > >
> > >
> >
> >
> > _______________________________________________
> > 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
> >
>
>
|
