[Rocks-Discuss] Changing the network ip addresses of the compute nodes

[Doll, Margaret Ann]

I have a ROCKS clusters using gridengine for the queues. It is using 5.0
ROCKS.

We are changing our networking at the school so that most of the computers
will have an ip address starting with 10.
This causes a problem with one of our clusters; we can't log into the
cluster computer on the 10.2 subnet.

Networking is suggesting:

You need to change the configuration of the eth0 interface so, at least for
now, its subnet mask is 255.255.255.0. I trust this is still a RedHat
server. If so, then the configuration file you would need to edit should be
located at /etc/sysconfig/network-scripts/ifcfg-eth0. I'm assuming here that
all the systems that exist on that cluster network are all addressed with
the first three octets as 10.255.255. If there are some where one of the
octets is a different 10 net then you need to let me know. Once you have
made the change you will need to perform a restart of network services on
TED or reboot TED entirely.

By changing the subnet mask to the one I suggested connectivity from all
Brown 10-nets should be enabled with the exception of anywhere else at Brown
that CIS may have deployed the 10.255.255.xx network. If some department is
issued that 10-net by CIS they would be unable to connect with TED since TED
still is using that network; hence my suggestion to readdress your cluster
network to a different RFC1918 network (selecting some portion of either
192.168.0.0/16 or 172.16.0.0/12). You may want to consult with NTG about
what address space you do chose so nothing else gets "black-holed."

What steps would I need to take to change the compute nodes using

0.255.255.216 compute-0-33.local compute-0-33 c0-33
10.255.255.217 compute-0-32.local compute-0-32 c0-32
10.255.255.218 compute-0-14.local compute-0-14 c0-14
10.255.255.219 compute-0-30.local compute-0-30 c0-30
10.255.255.237 compute-0-21.local compute-0-21 c0-21
10.255.255.238 compute-0-20.local compute-0-20 c0-20
10.255.255.239 compute-0-19.local compute-0-19 c0-19
10.255.255.240 compute-0-18.local compute-0-18 c0-18
10.255.255.241 compute-0-17.local compute-0-17 c0-17
10.255.255.242 compute-0-16.local compute-0-16 c0-16
10.255.255.243 compute-0-22.local compute-0-22 c0-22
10.255.255.244 compute-0-31.local compute-0-31 c0-31
10.255.255.245 compute-0-13.local compute-0-13 c0-13
10.255.255.246 compute-0-12.local compute-0-12 c0-12
10.255.255.247 compute-0-11.local compute-0-11 c0-11
10.255.255.248 compute-0-10.local compute-0-10 c0-10
10.255.255.249 compute-0-7.local compute-0-7 c0-7
10.255.255.250 compute-0-6.local compute-0-6 c0-6
10.255.255.251 compute-0-3.local compute-0-3 c0-3
10.255.255.252 compute-0-2.local compute-0-2 c0-2
10.255.255.253 compute-0-1.local compute-0-1 c0-1
10.255.255.254 compute-0-0.local compute-0-0 c0-0

to the 192.168.0.0/16 scheme or the 172.16.0.0/12 scheme. Our ROCKS
cluster ROCK 4.2 used the 172.16.0.0/12 scheme automatically.

Thanks for your help.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20110707/3982db36/attachment.html

[LaoTsao]

http://www.rocksclusters.org/roll-documentation/base/5.0/x3167.html
rocks set host interface ip compute-0-1 eth0 new ipaddr

Sent from my iPad
Hung-Sheng Tsao ( LaoTsao) Ph.D

[LaoTsao]

One also need to change pxe server and dhcp server eth0 address
Rocks dump should provide you some info
Regards

Sent from my iPad
Hung-Sheng Tsao ( LaoTsao) Ph.D