Thread: compute node install failures due to LVM on compute nodes

Greetings,

I had a major issue with a PoC install of the Red Hat HPC software, which you all know is built off of kusu.

After I debugged some of the various issues with the TFTP booting, and other required services, I had major issues everytime I tried to provision a compute node.

We were given a number of re-purposed x2100s to due for our testing. Well they all had LVM partitions/volumes on them, and the installer was failing. Apparently, the kickstart file has issues destroying the LVM volumes.

So I completely iced the compute node with GParted, and then the installer ran perfectly. I just wanted to throw this out here in case anyone has an issue like this.

PS: can I just manually edit the kickstart file to add in directives to remove ALL LVM/etc from new nodes?

Also, I didn't see anything like this mentioned in any documentation, but I know the internet is a large place. Most of my reference I am getting from here. Is there anywhere I can more documentation for kusu and such, so I might avoid mistakes like this in the future?

Hi, sorry about the late response. This was NOT a drive-by posting. Just with the holiday and a ton of other tasks, I've neglected working with this software.

I'll send you the requested information when I duplicate it again.