|
On Sun, 2010-06-27 at 13:25 -0400, William Ottley wrote:
> Any help in this matter would be greatly appreciated.
Seeing the config for your LVS would help.
That said...
If you're using persistence - all requests from a single client will go
to a single webserver.
If your client app, and the Apache webserver, are configured to use
keepalives - it's quite likely all (or a large proportion) of the
requests will go to a single webserver; this is because the connections
don't get closed until the complete client session has ended.
What you need to do is to run the test again, over a longer period, and
ensure that the balance of requests is 50/50 over the longer duration.
Use the following to monitor this:
watch -n1 -- "ipvsadm -Ln; echo; ipvsadm -Ln --rate"
If the spread isn't 50/50 then configure your LVS to make it that way,
and make sure the client/server pair are configured in as basic a way as
possible.
Also note that you could be filling up network segments, interfaces, or
buffers.
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
|
