Is MPI factory installed on Octane dual CPU

[Alexis Cousein]

Hello,
I have an Octane machine, dual processor, came from SGI w/ O/S
installed, & w/
the dual CPU's. I would assume then that sgi's MPI is installed,

Do not assume. MPT (includes mpi) is on
<URL:http://www.sgi.com/products/evaluation/>

The command for running MPI programs is mpirun, not mpi.

[Adam Smith]

Hello,

I am experiencing a partial network failure on my network which consists
of several SGI's, PC's, FreeBSD's and Mac's, but only the SGI's are
affected. They cannot ping or be pinged in the LAN. On a reboot they are
operable for about 15 minutes before going down.

Hope someone with a solution sees this

Thanks

-- Adam --

[Walter Roberson]

Post by Adam Smith
I am experiencing a partial network failure on my network which consists
of several SGI's, PC's, FreeBSD's and Mac's, but only the SGI's are
affected. They cannot ping or be pinged in the LAN. On a reboot they are
operable for about 15 minutes before going down.

Are they on a different subnet? If so then that sounds as if they might
not have a default route.

[Adam Smith]

Post by Walter Roberson

Post by Adam Smith
I am experiencing a partial network failure on my network which consists
of several SGI's, PC's, FreeBSD's and Mac's, but only the SGI's are
affected. They cannot ping or be pinged in the LAN. On a reboot they are
operable for about 15 minutes before going down.

Are they on a different subnet? If so then that sounds as if they might
not have a default route.

This is what we addressed first, the ifconfig for each machine was and
did set an explicit default route in
/etc/config/static-route.options
=> Added explicit default net to route
line added => route -n add -net default 192.168.2.1 -hopcount 1
----------------------------------

machone 24# ifconfig -a
ec0: flags=400c43<UP,BROADCAST,RUNNING,FILTMULTI,MULTICAST,IPALIAS>
inet 192.168.2.5 netmask 0xffffff00 broadcast 192.168.2.255
lo0: flags=1849<UP,LOOPBACK,RUNNING,MULTICAST,CKSUM>
inet 127.0.0.1 netmask 0xff000000

and ----------------

machtwo 24# ifconfig -a
ef0:
flags=8415c43<UP,BROADCAST,RUNNING,FILTMULTI,MULTICAST,CKSUM,DRVRLOCK,LINK0,IPALIAS,IPV6>
inet 192.168.2.14 netmask 0xffffff00 broadcast 192.168.2.255
lo0: flags=8001849<UP,LOOPBACK,RUNNING,MULTICAST,CKSUM,IPV6>
inet 127.0.0.1 netmask 0xff000000

_________________________________________________________
netstat -in
machone =>

Name Mtu Network Address Ipkts Ierrs Opkts Oerrs
Coll
ec0 1500 192.168.2 192.168.2.5 313 0 324 0
0
lo0 8304 127 127.0.0.1 17845 0 17845 0
0

and ----------------
machtwo =>

Name Mtu Network Address Ipkts Ierrs Opkts
Oerrs Coll
ef0 1500 192.168.2 192.168.2.14 204 0 653
0 0
lo0 32992 127 127.0.0.1 16762 0 16762
0 0
----------------------------------

However I have just noticed that the
eoe.sw.tcp package is missing from the installation.
aren't these supposed to be present for the
IRIX 6.5 6.5.16f
IRIX64 6.5 6.5.23m
distro.

Thanks

[Gary Heston]

Hello,
I am experiencing a partial network failure on my network which consists
of several SGI's, PC's, FreeBSD's and Mac's, but only the SGI's are
affected. They cannot ping or be pinged in the LAN. On a reboot they are
operable for about 15 minutes before going down.
Hope someone with a solution sees this

Once the problem occurs, can they ping each other?

Can you be mre descriptive about the topology--is this a switched LAN,
several LANs routed, are the SGIs on their own network, is there anything
else on the LAN with them?

Have you compared ARP tables before/after the problem occurs? That could
show a "rogue gateway" with the same address as the real one. I've run
into that before; causes wierd problems.

Are network configurations static or via DHCP?


Gary

[Adam Smith]

Post by Gary Heston

Hello,
I am experiencing a partial network failure on my network which consists
of several SGI's, PC's, FreeBSD's and Mac's, but only the SGI's are
affected. They cannot ping or be pinged in the LAN. On a reboot they are
operable for about 15 minutes before going down.
Hope someone with a solution sees this

Once the problem occurs, can they ping each other?

No they can't
Usually => "No route to host" but PC's and FreeBSD can ping each other

Post by Gary Heston
Can you be mre descriptive about the topology--is this a switched LAN,
several LANs routed, are the SGIs on their own network, is there anything
else on the LAN with them?

Topology - DSL ==> Modem ==> Router ===> 16 port Hub/Switch
Gateway is configured in Router as 192.168.2.1
Switched
Single LAN
Mixed network 2 SGI's, 1 FreeBSD, 1 Mac and 3 PC's

Post by Gary Heston
Have you compared ARP tables before/after the problem occurs? That could
show a "rogue gateway" with the same address as the real one.
I've run
into that before; causes wierd problems.

Weird indeed, the fact that you have experienced this before is a bit
reassuring to me, in that I am no longer doubting my sanity

mach2 4% arp -n mach1
mach1 (192.168.2.5 --no entry
mach2 5% arp -n mach2
mach2 (192.168.2.14 --no entry

Post by Gary Heston
Are network configurations static or via DHCP?

Static IP
mach2 9% traceroute mach1
traceroute to mach1 (192.168.2.5), 30 hops max, 60 byte packets
1 traceroute: send_ipv4: sendto: No route to host
mach2 10%

From mach1

traceroute to mach2 (192.168.2.14), 30 hops max, 40 byte packets
sendto: No route to host


rup => shows machines on network
ifconfig -a output

machtwo ==>
ifconfig -a
ef0:
flags=8415c43<UP,BROADCAST,RUNNING,FILTMULTI,MULTICAST,CKSUM,DRVRLOCK,LINK0,IPALIAS,IPV6>
inet 192.168.2.14 netmask 0xffffff00 broadcast 192.168.2.255
lo0: flags=8001849<UP,LOOPBACK,RUNNING,MULTICAST,CKSUM,IPV6>
inet 127.0.0.1 netmask 0xff000000

machone ==>
ifconfig -a
ef0: flags=400c43<UP,BROADCAST,RUNNING,FILTMULTI,MULTICAST,IPALIAS>
inet 192.168.2.5 netmask 0xffffff00 broadcast 192.168.2.255
lo0: flags=1849<UP,LOOPBACK,RUNNING,MULTICAST,CKSUM>
inet 127.0.0.1 netmask 0xff000000


Thanks!

Post by Gary Heston
Gary

[Adam Smith]

Hello,

I've run into that before; causes wierd problems.

How did you solve your problem ?

Gary

[Gary Heston]

Post by Adam Smith

Hello,

[ rogue gateway ]

Post by Adam Smith

I've run into that before; causes wierd problems.

How did you solve your problem ?

Check the arp table every few minutes and if the MAC address
for the gateway changes when things stop working, note the
new MAC address, then start looking for a piece of equipment
that has that MAC.

In your situation, get before and after output from ifconfig
and the route tables on the affected systems and compare them;
perhaps something will show up.


Gary

[Adam Smith]

Post by Gary Heston

Post by Adam Smith

Hello,

[ rogue gateway ]

Post by Adam Smith

I've run into that before; causes wierd problems.

How did you solve your problem ?

Check the arp table every few minutes and if the MAC address
for the gateway changes when things stop working, note the
new MAC address, then start looking for a piece of equipment
that has that MAC.
In your situation, get before and after output from ifconfig
and the route tables on the affected systems and compare them;
perhaps something will show up.
Gary

Just decided to rebuild the network and settings, that worked, NFS
drives and ping capabilities appeared. One problem appeared however,
that I think is minor but the man pages aren't too clear. Followed
instructions to use command 'proclaim' after set up and promptly broke
one of the NFS connections. 'proclaim' does an automatic DHCP config, I
use static IP throughout my system and can't figure out how to reverse
this. Tried dropping the lease but this did not seem to help.

Any suggestions?

-- Adam --

[Gary Heston]

In article <***@speakeasy.net>,
Adam Smith <***@econ.com> wrote:
[ ... ]

Post by Adam Smith
Just decided to rebuild the network and settings, that worked, NFS
drives and ping capabilities appeared. One problem appeared however,
that I think is minor but the man pages aren't too clear. Followed
instructions to use command 'proclaim' after set up and promptly broke
one of the NFS connections. 'proclaim' does an automatic DHCP config, I
use static IP throughout my system and can't figure out how to reverse
this. Tried dropping the lease but this did not seem to help.
Any suggestions?

Based upon a quick look at:

http://techpubs.sgi.com/library/tpl/cgi-bin/getdoc.cgi?coll=0650&db=man&fname=/usr/share/catman/u_man/cat1/proclaim.z

you may need to change the NVRAM netaddr variable. Proclaim updates it,
and you can probably determine what it should contain by looking at your
other SGI system.


Gary