During the installation of ESX all the standard Linux devices are available as ESX is not running at that time. The DL385 will install from a kickstart file on a floppy. Personally I like to PXEboot my servers and install over the network, but that requires a PXEboot, DHCP, and HTTPD or NFS servers to get going. This way I do not need to worry about a floppy device or even being in the data center since the DL machines have built in ILO which allows you to specify a virtual floppy or cdrom device, regardless of the hardware in the server.
But in answer to your question, that should work.
1 person found this helpful
To reintegrate the boot.msg and various files back into the image you need to extract the initrd image, reconfigure it, and recompress it. See basic instructions at http://www.vmprofessional.com/index.php?content=kickstart_2 from Dominic.
P.S the virtual media license costs money on the DL385 and frankly isn't very quick. A usb floppy should work if you get the drivers loaded and is a quick way to test changes in your config without running the above process everytime. For debugging and test it's easiest to start with the floppy and cdrom. There are still issues with some broadcoms and cisco switches that confound PXE booting.
I must agree with Jae, if you do not have the Advanced Option License for the iLO, you can not use the remote media, but if you do, it certainly saves some time. I Have done installs using it across the nation. Just be sure to set your timeouts to the maximum.
There are many ways to install a Linux machine and at the install stage of ESX, you can treat it like a standard Linux machine.
Thank you for your responses. I will say I'm as good as a newbie when it comes to UNIX-like environments, but found a thread on here that answered by iLO virtual floppy setup questions. Got that to work, no problems there, so that removes any quirks that could arise from the USB floppy, although I think I saw comments on that somewhere too ... should it come to that.
We have PXE servers here and I've since been informed the vendor we'll be using to blast the servers with ESX do also (it wouldn't have been very encouraging to us if they didn't!). Never played with PXE, so hopefully it's intuitive having it offer ESX to the LAN.
Question regarding the floppy solution: Is the name of the KS file the scripted install references fixed to KS.CFG? Ideally I'd be able to have all the KS files in one place and be able to select at scripted install trigger time.
If you are using the iLO floppy solution, then the kickstart file on the floppy must be named 'ks.cfg', note there are no capital letters. Therefore you would have one floppy per different ks.cfg file.
If you are using the PXEboot method then that depends on how the PXEboot server is setup. Each setup has their own naming convention for the kickstart file. Actually, in this case the name is irrelevant and you can name them after machines or even revisions.
OK - I need more assistance if someone can clue me in. I have it to where I've got my ks.cfg on a floppy disk presented over iLo. When I enter the "esx ks=floppy method=cdrom" string at the initial ESX Server 3 CD bootup prompt, the light to the floppy drive does light up ... but in progressing beyond that point, the installation process doesn't grab any info from the ks.cfg file: I get the prompts and default partition layout (we have 2 RAIDs and so reconfig from the norm), plus a blank NIC settings screen.
I called tech support and got the response that the required drivers probably aren't available at the point where the esx line is typed. I was pointed in another direction though:
esx ks= method=cdrom
I did this and believe I now have ESX reading the ks file ... and now the following error is presented:
"raise ValueError: partition command requires one anonymous argument"
install exited abnormally.
The following is the partition info from the ks file:
\# Clear Partitions
clearpart --all --initlabel --drives=cciss/c0d0,cciss/c0d1
part /boot --fstype ext3 --size 199 --ondisk cciss/c0d0
part / --fstype ext3 --size 10001 --ondisk cciss/c0d0
part swap --size 1602 --ondisk cciss/c0d0
part --fstype vmfs3 --size 102000 --grow --ondisk cciss/c0d0
part None --fstype vmkcore --size 110 --ondisk cciss/c0d0
part /var/log --fstype ext3 --size 1992 --ondisk cciss/c0d0
part /vmimages --fstype ext3 --size 25000 --ondisk cciss/c0d0
part --fstype vmfs3 --size 559922 --grow --ondisk cciss/c0d1
anything wrong with this??
Again - TIA for any input
What was wrong was the absence of None for the vmfs3 part lines. Needs it.
I get the same error and i have "None" infront of vmfs3
part /boot --fstype ext3 --size 102 --ondisk cciss/c0d0
part / --fstype ext3 --size 5000 --ondisk cciss/c0d0
part swap --size 1600 --ondisk cciss/c0d0
part None --fstype vmfs3 --size 10000 --grow --ondisk cciss/c0d0
part None --fstype vmkcore --size 100 --ondisk cciss/c0d0
part /var --fstype ext3 --size 5000 --ondisk cciss/c0d0
part /var/log --fstype ext3 --size 5000 --ondisk cciss/c0d0
part /home --fstype ext3 --size 2000 --ondisk cciss/c0d0
part /tmp --fstype ext3 --size 2000 --ondisk cciss/c0d0
part /vmimages --fstype ext3 --size 10000 --ondisk cciss/c0d0