|
hi there,
i happene to see ur query in LVS formus and we believe that we have an
answer for you question , i am samrat patel frm india i am a graduate in
computer sci and engineering and also a researcher in papralle processing
and cluster computing .. well we have already tried this experiment but we
tried this on the my private LAN of my ISP provider ...well still there are
some loopholes in this still we managed to work it out well ...also you will
have to consider various issues such as the OS both on the clients and the
server ..? well answer to you query is possible ...
plz to contact us .. we are eager to help you ... we will have to understand
each other well
eagerly awaiting your response
thanking you
samrat
From: lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
Reply-To: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
To: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Subject: lvs-users Digest, Vol 43, Issue 13
Date: Tue, 15 Aug 2006 12:00:23 +0200 (CEST)
Send lvs-users mailing list submissions to
lvs-users@xxxxxxxxxxxxxxxxxxxxxx
To subscribe or unsubscribe via the World Wide Web, visit
http://www.in-addr.de/mailman/listinfo/lvs-users
or, via email, send a message with subject or body 'help' to
lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
You can reach the person managing the list at
lvs-users-owner@xxxxxxxxxxxxxxxxxxxxxx
When replying, please edit your Subject line so it is more specific
than "Re: Contents of lvs-users digest..."
Today's Topics:
1. need some insight and direction (cschoon@xxxxxxxxxx)
2. Re: need some insight and direction (Jason Martin)
3. Lost packets and dead/warntime (Sebastian Vieira)
4. Re: Lost packets and dead/warntime (Graeme Fowler)
From: cschoon@xxxxxxxxxx
Reply-To: "LinuxVirtualServer.org users mailing
list."<lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
To: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Subject: need some insight and direction
Date: Mon, 14 Aug 2006 15:01:41 -0400
I am a consultant for a large law firm in Orlando, Fl. I have been
asked to do some research into Linux Clustering. to be specific, the
vision is to setup a linux cluster, run VMWare on the cluster, then
run Microsoft / Novell servers in VMWare.
If that is possible, then is it possible to add all 200+ workstations
to the cluster and run the desktop OS (Win XP / Win 98) on top of
that? Thus creating a cluster in the neighborhood of 243 nodes.
From: Jason Martin <jhmartin@xxxxxxxx>
Reply-To: "LinuxVirtualServer.org users mailing
list."<lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
To: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Subject: Re: need some insight and direction
Date: Mon, 14 Aug 2006 13:18:13 -0700
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Mon, Aug 14, 2006 at 03:01:41PM -0400, cschoon@xxxxxxxxxx wrote:
> I am a consultant for a large law firm in Orlando, Fl. I have been
> asked to do some research into Linux Clustering. to be specific, the
> vision is to setup a linux cluster, run VMWare on the cluster, then
> run Microsoft / Novell servers in VMWare.
>
> If that is possible, then is it possible to add all 200+ workstations
> to the cluster and run the desktop OS (Win XP / Win 98) on top of
> that? Thus creating a cluster in the neighborhood of 243 nodes.
Lets back up a second.
VMware is a tool that allows you to run a set of virtual
machines on one physical machine. VMware ESX has the concept of
a 'cluster' whereby VMs will automatically migrate between
physical machines to balance load, or automatically reboot onto
a working node if the primary as failed. This feature requires
that all of the servers involved used a shared storage device.
For server applications, you could certainly run the servers
inside a VMWare VM.
Assuming you had multiple VMs performing a given webhosting
task, you could use Linux LVS to balance the incoming traffic
across them.
However, Linux LVS doesn't have the concept of 'adding machines
to a cluster' in the same sense that a NUMA supercomputer can add
nodes and treat them as a single large image. The physical
machines are still disparate.
I don't know that it makes any sense to have your desktops /
workstations run VMWare and then run Windows on top of it. It
would make more sense to have your desktops run a cut down
Windows natively, then RDP into a VM running on dedicated VMWare
servers.
Perhaps you could give us a better idea of what you are trying
to accomplish?
- -Jason Martin
- --
Taglines can be more interesting than messages!
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)
Comment: --no-verbose
iD8DBQFE4NqFeDShAkRw0YoRAldmAKDNhDm9PyyWSumr9McSB0ELUvRH3wCeP2nT
DeHu62ztjaLa2kqHd/woGcE=
=KPgZ
-----END PGP SIGNATURE-----
From: "Sebastian Vieira" <sebvieira@xxxxxxxxx>
Reply-To: "LinuxVirtualServer.org users mailing
list."<lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
To: "LinuxVirtualServer.org users mailing
list."<lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Lost packets and dead/warntime
Date: Tue, 15 Aug 2006 09:11:17 +0200
Hi,
I'm moving 2 LVS's from VMWare to physical boxes, but strangely enough i
get
better performance with failovers on vmware, than i get on the physical
machines. They are now setup with 3 nics; eth0 and eth1 are bonded (trés
cool) and are used for all the client/server connections, and eth2 is used
for the heartbeat and sync daemons. Kernel used is 2.6.16 (gentoo-r9).
On vmware i rarely had lost packets, but on the physical boxes i get about
3-5 lost packets each 24h. My settings from ha.cf:
keepalive 50ms
deadtime 3
warntime 1
initdead 120
That's one thing. The other is when i used to do a failover on vmware
almost
all connections were transferred, and now less than 10% :|
Anyone with a tip?
Sebastian
From: Graeme Fowler <graeme@xxxxxxxxxxx>
Reply-To: "LinuxVirtualServer.org users mailing
list."<lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
To: "LinuxVirtualServer.org users mailing
list."<lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: Lost packets and dead/warntime
Date: Tue, 15 Aug 2006 09:20:56 +0100
On 15/08/2006 08:11, Sebastian Vieira wrote:
<snip>
Anyone with a tip?
Stop your eth2 NICs from autonegotiating with the appropriate driver
options in modules.conf, modprobe.conf, /etc/modules or whatever your
distro uses. If they're 100 meg cards, force them both to 100BaseTX-FD; for
gig cards force them to 1000BaseTX-FD.
It sounds to me like you've got both ends of the link continually
renegotiating, which will mean a certain amount of errors - although they
may retransmit at L2, it might mean you get L3 timeouts on occasion.
Inside your VMs you won't have this problem (obviously) unless you're
running on two VMWare host servers with a corresponding physical problem.
You can fiddle with the card settings by using mii-tool in real time.
Graeme
_______________________________________________
lvs-users mailing list
lvs-users@xxxxxxxxxxxxxxxxxxxxxx
http://www.in-addr.de/mailman/listinfo/lvs-users
_________________________________________________________________
Connect with your friends who use Yahoo! Messenger with Voice. Click!
http://www.msnspecials.in/wlmyahoo/index.asp
|
