dual quad-core Nehalem node!
April 20, 2010by Wm. Josiah Erikson (wjens)
Very exciting… Lee has just funded us buying a Dell PowerEdge R410 with 24GB of RAM and dual quad-core 2.26Ghz Nehalems. This will show up for most intents and purposes as a 16-processor node. Should be fun. Fighting with kernels for the install – had to get a newer pxeboot kernel to support the BCM5716 and then put the newer kernel into the distro as well – we’ll see if this screws things up with the distro – I just took the packages from newer CentOS and ROCKS 5.3 (rocks-boot and kernel-*) without actually upgrading to ROCKS 5.3 – we’re still running 5.1. Cross your fingers!
April 20th, 2010 at 3:29 pm
It appears to have worked! Hooray. Just grabbing the kernel packages from CentOS 5.4 and the rocks-boot package from ROCKS 5.3 seems to have done the trick. Most excellent. Now I have to see if I can figure out some weird way to make those new Dell rails work in threaded-hole racks.
April 20th, 2010 at 3:31 pm
Now we see if I can fix that problem where the nodes won’t reinstall from the local boot kernel… I think I might understand enough now to do that.
Man, I can’t wait until this is a single-arch cluster again! Hey Lee, got some more money for a new frontend node? 🙂
April 20th, 2010 at 4:46 pm
I think I got the default reinstall on poweroff working again… neat! No more touching every single node every time the power goes out!
April 21st, 2010 at 1:37 pm
Why are some of the nodes (compute-2-3 and 2-8 in particular) refusing to reinstall with “cannot retreive repository metadata” while other nodes are reinstalling fine? I’m pretty sure a few of the nodes that failed before worked the second time I tried, which makes even less sense. This started happening after I rebuilt the i386 distro so that the i386 nodes would reinstall. Maybe after compute-0-2 is done reinstalling, I’ll rebuild the x86_64 distro again and see if compute-2-3 and 2-8 suddenly behave themselves. Again, I can’t wait to rebuild this cluster x86_64 only!
April 22nd, 2010 at 9:13 am
Huh. This appears to be a side effect of the new pxeboot kernel from rocks-boot from ROCKS 5.3. I’m not sure if it would do it sometimes after installed, too. I wonder if I installed ROCKS 5.3 whether I would see this problem all the time… that would suck. But why? It’s so random. I’ll have to do some more testing… does it happen to the other nodes, or just the blades (compute-2-x)?