Re: DHCP in openMosix
"Ian Latter" <[email protected]>
| Newsgroups | gmane.linux.cluster.openmosix.general |
|---|---|
| Message-ID | <[email protected]> |
Dude,
There are two possibilities here;
1. NETWORK ISSUE
So long as the following hold true;
- all of your openMosix machines have only one NIC
- all of your openMosix machines are connected to the
same LAN or VLAN (i.e. none are seperated by routers)
.. then your problem is a network configuration issue.
In your environment, you have;
- Vader at 192.168.4.144/255.255.255.0
- A second machine at 192.168.2.14/(what mask?)
- And other machines per DHCP allocation
Vader is unable to see the second machnie, and
probably the DHCP allocated machines as 192.168.4.144/
255.255.255.0 must route to something on the
192.168.4.x network to get to 192.168.2.14.
Without a router, and while you are using different
networks with different masks, then neither auto-
discovery nor static maps will work.
A good network test for an unfiltered environment; can
you ping from each cluster node to each other cluster
node?
2. OPENMOSIX ISSUE
If, however, your network is setup in a more typical
manner;
- all of your openMosix machines have only one NIC
- all of your openMosix machines are connected to
LANs or VLANs that are routed (and not filtered)
.. then your problem is an openMosix configuration
issue.
Try disabling auto-discovery (in this routed environment)
and use the following openMosix Map file on *all* cluster
machines;
# Static openMosix configuration
# ==============================
#
# Each line in this file should contain 3 fields, statically mapping
# IP addresses to openMosix node-numbers:
#
# 1) The first openMosix node-number in this range.
# 2) The IP address of the above node (or node-name from /etc/hosts).
# 3) The number of nodes in this range.
#
# Note: If you don't create a valid IP <-> node-number mapping, the
# autodiscovery daemon will be started, automatically assigning
# node-numbers to all visible openMosix machines.
#
# Example: 10 machines with IP addresses 192.168.1.50 - 192.168.1.59
# which will have openMosix node-numbers 1-10:
#
# 1 192.168.1.50 10
#
# MOSIX-# IP number-of-nodes
# ============================
1 192.168.2.1 255
256 192.168.4.1 255
Note that this assumes that you DHCP allocation is
going to populate either the 192.168.2.x network or
the 192.168.4.x network. If your DHCP machines
are in another network, say 192.168.3.x, then add
the following line to the above openmosix.map file;
512 192.168.3.1 255
Hope that helps,
----- Original Message -----
>From: "Murali P" <[email protected]>
>To: "openMosix General List" <openmosix-general-5NWGOfrQmneRv+LV9MX5uuRhgaa4a2kL@public.gmane.org
net>
>Subject: Re: [openMosix-general] DHCP in openMosix
>Date: Sat, 28 Jan 2006 12:37:48 +0530
>
> On 26 Jan 2006 15:32:22 +0100, Moreno Baricevic <baro@democritos.
it> wrote:
> >
> > Hi Murali,
> >
> > we need some more info. Could you tell us something about your
> > openmosix.map and your openmosix.config? Are you using omdiscd or
a
> > static oM map? Do you have one or more NICs on your nodes? Are you
sure
> > you are running the same oM kernel everywhere? Did you start
openmosix?
> > Any warning or error message when you start the oM service? Does
the
> > command "setpe -c -f /etc/openmosix.map" say anything useful?
> >
> > Ciao
> >
> > Moreno
> >
> >
> >
>
> On 1/26/06, John Floren <[email protected]> wrote:
> >
> > Okay, let's try to figure this out. You say you can ping and ssh the
> > hosts, but they are not "recognizing" each other? By that do you mean
> > that openMosix isn't finding any other hosts? Did you configure an
> > openmosix.map file? Or are you trying to do auto-discovery?
> > I never had much luck it auto-discovery, but that was two years ago. A
> > little more information would be helpful to get your problem fixed.
> > Thanks
> >
> > John
> >
>
> I had used openmosix-kernel-2.4.26-openmosix1.i686.rpm for installing
> openMosix.
>
> I dont have access to one of the machines i had used now. this is the
> configuration in the second machine.
>
>
> [root@vader openmosix-downloads]# cat /etc/openmosix/openmosix.
config
> #
> # This file can be used to change the default behaviour of the
> # openMosix startup-script.
> #
>
> # Force autodiscovery-daemon to start, even with a valid .map-file
> # The following line uncommented by Murali to force autodiscovery
> AUTODISC=1
>
> # Specify which network interface the autodiscovery-daemon should
listen to
> # AUTODISCIF=eth1
>
> # Set the values of /proc/hpc/admin/overheads
> # OVERHEADS=
>
> # Set the values of /proc/hpc/admin/mfscosts
> # MFSCOSTS=
>
> # Set the openMosix node-id of this node
> # MYOMID=
>
> # Set maximum number of gateways between openMosix nodes (see
man setpe)
> # MOSGATES=
>
> # Pass the following value to mosctl setspeed when starting the node
> # man mosctl for details
> # NODESPEED=10000
>
> # Processes are allowed to migrate to other nodes.
> # MIGRATE=yes
>
> # Allow guest processes to arrive.
> # BLOCK=no
>
> # Don't use MFS
> # MFS=yes
>
> I hav used the same config file on both machines.
>
> First I had tried the defualt settinmg. Then I forced autodiscovery as well.
>
>
> [root@vader openmosix-downloads]# cat /etc/openmosix.map
> # Static openMosix configuration
> # ==============================
> #
> # Each line in this file should contain 3 fields, statically mapping
> # IP addresses to openMosix node-numbers:
> #
> # 1) The first openMosix node-number in this range.
> # 2) The IP address of the above node (or node-name from /etc/hosts).
> # 3) The number of nodes in this range.
> #
> # Note: If you don't create a valid IP <-> node-number mapping, the
> # autodiscovery daemon will be started, automatically assigning
> # node-numbers to all visible openMosix machines.
> #
> # Example: 10 machines with IP addresses 192.168.1.50 - 192.168.1.59
> # which will have openMosix node-numbers 1-10:
> #
> # 1 192.168.1.50 10
> #
> # MOSIX-# IP number-of-nodes
> # ============================
> 526 192.168.2.14 1
>
> I had tried giving MOSIX-# as 1 first. then i tried 526 as well. IP of vader
> is 192.168.4.144. I have tried commenting the last line as well.
> network mask is 255.255.255.0
>
> When i deploy finally, ALL the IP addresses of the machins othr than
> 192.168.2.14 will be got from DHCP only.
>
> All my machines are having one NIC only.
>
> [root@vader root]# service openmosix restart
> openMosix: Using map file /etc/openmosix.map
> openMosix: WARN: Invalid configuration in map-file /etc/openmosix.map
> openMosix: Falling back to autodiscovery mode using /sbin/omdiscd
> Stopping openMosix...
> openMosix: All remote processes were expelled and no further remote
> processes accepted.
> Initializing openMosix...
> openMosix configuration was successful :)
>
> [root@vader root]# tail -7 /var/log/messages
> ...
> Feb 1 04:32:31 vader kernel: openMosix: All guest processes evacuated
and
> no further guests allowed in
> Feb 1 04:32:31 vader kernel: openMosix configuration disabled
> Feb 1 04:32:31 vader omdiscd[1208]: Exiting on SIGINT.
> Feb 1 04:32:31 vader kernel: openMosix: guest-processes now accepted
again.
> Feb 1 04:32:31 vader kernel: openMosix configuration changed: This is
> openMosix #1168 (of 1 configured)
> Feb 1 04:32:31 vader kernel: openMosix #1168 is at IP address 192.168.
4.144
> Feb 1 04:32:31 vader omdiscd[1263]: Notified kernel to activate
openMosix
>
> Even when the other machine was connected, the /var/log/messages
output was
> same only.
>
>
>
> --
> see not what comes to u.
> see what comes out of u.
>
--
Ian Latter
Late night coder ..
http://midnightcode.org/
-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems? Stop! Download the new AJAX search engine that makes
searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642