We have an existing iDataPlex cluster, and we're adding 2 more racks. One 
of the systems in the new racks will be a service node, but the new 
hardware, including the new SN, is on a different subnet. The compute 
nodes in the new racks will be netbooting, and their images will need to 
be built on the SN since the MN is running an older version of RH than 
they want on the new hardware. They don't want the SN running a DHCP 
server if we can avoid it. All we're looking to have the MN do is handle 
DHCP and ipmi (r-tools, console). That said, is it safe to say a minimal 
list of ports to open up on the firewall between the two racks would be as 
follows:

xcatd
xcati
dhcp
dhcpc
ipmi
pxe
postgresql (this wasn't listed on the wiki)
conserver 
rpc-mount

Have I missed anything?

The plan is that the compute nodes will netboot off the SN, but the 
initial PXE/DHCP communication will be handled by the MN. Admin tools like 
rpower & rcons will be handled on the MN. We're not doing node discovery 
(MACs have been manually collected) and DNS is handled by their servers.

Regards,

Christian D. Caruthers
Senior Consultant - System x Linux HPC


Mobile: 1-757-289-9872 | Phone: 1-804-327-4559
E-mail: [email protected]
Find me on:  
"A common mistake that people make when trying to design something 
completely foolproof is to underestimate the ingenuity of complete fools." 
- Douglas Adams


9201 Arboretum Pkwy
Richmond, VA 23236-5402
United States

<<image/jpeg>>

<<image/jpeg>>

<<image/gif>>

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

Reply via email to