VMware Cloud Community
Schorschi
Expert
Expert

Deploy ESX 4 via Fedora Cobbler?

Ok, this is way way off the RADAR, but anyone using Cobbler? To try to deploy ESX 4? This should work, given that ESX 4 has an vmlinuz and initrd.img file. I know my Cobbler setup is good, I can for example provision Fedora 10 without any issues. The interesting thing is, ESX 4 vmlinuz downloads, so does initrd.img, but when the bootstrap executes, it drops in the the recovery shell saying... "Failed to load VMkernel" and then the session freezes. Cobbler does not officially support ESX but still, this is a puzzle.

0 Kudos
5 Replies
admin
Immortal
Immortal

I know Cobbler supports VMware Server. Did you try ESXi 4 also?

VMware-GSS,

VMware Communities User Moderator

0 Kudos
Schorschi
Expert
Expert

Trying to do bare-metal deployment of ESX 4 classic, not VMs, not sure I made that point clear above. Figured I would try ESX 4 classic before I tried to tackle ESX 4i.

0 Kudos
Schorschi
Expert
Expert

I also tried BladeLogic 7.5.x which we use as well, and it has the exact same issue, in the exact same place. The vmlinuz and initrd.img files are served up by PXE/TFTP just fine, everything appears fine, only to have bootstrap crash saying VMKernel could not be loaded. So this appears to be an issue with initrd.img file its self? I am not passing a ks= parameter to specify network information, at boot but attempting to use just a url --url command option in kickstart. Figure at worst the VMware ESX OS installer would grip about not finding a kickstart file, but I am not even getting to that point, where it is looking for a kickstart configuration file. I posted a screen print of the error.

0 Kudos
Struer
Contributor
Contributor

As you can see in your error msg. it tries to access tty, so if you make a debug log like I've done for my PXE installlation, you should get past that point.

append ksdevice=vmnic0 lang= text initrd=initrd.img vmkopts=debugLogToSerial:1 mem=512M ks=ftp://...

0 Kudos
Schorschi
Expert
Expert

Actually, it appears it was the mem=512 parameter I had missed adding or omitted initially. So the vmkernel did not have the resources required to spin up.

0 Kudos