I have a similar problem, with the DL 160 G5. The installer says it can't find any supported NICs

(5722s: RH EL 5 - 64 says it uses the tg3 driver, and its supposed to be in 3.5, so, don't get it).

thanks in advance,

Matthew

With ESX, virtual machines only can use virtual SCSI disks and XP doesn't include the drivers for that. See this thread on how to get around that - http://communities.vmware.com/message/819716?tstart=0.

I'm also trying to evaluate ESX 3.5 on two DL180 G5s. I'm using local SATA drives and have tried using the service console to edit /etc/vmware/pciid/ata_piix.xml to include the DL180-specific device ID (based on instructions SATA Support Clear as Mud!?). I added this to the Intel vendor section (8086).

<device id="2920:103c:31fe">
<vmware label="scsi">
<driver>ata_piix</driver>
<vmware>
<name>HP DL180 G5 integrated 4 port SATA</name>
</device>

I then ran esxcfg-pciid and then esxcfg-boot -p but even after boot I get errors mounting the root filesystem. I am able to confirm that the proper PCI IDs have been added to the initrd's config files.

I have SATA RAID mode disabled and the installer sees the devices just fine. After some research it seems that this hardware is not "officially" supported but it doesn't seem to be different from other working ICH9 controllers (different device ID perhaps but the service console works using the ata_piix driver). There doesn't seem to be a way in the BIOS to select native or legacy SATA modes.

Per the ESX 3.5 whitebox table it seems like this should work... am I missing something or should I get new hardware? I also tried 3i but post-install I had the same error mounting the root filesystem (but haven't tried the PCIID hack yet).

I've seen some other posts mention using an HTTP-based install (Building a test server / white box - post your builds here), so I'll give this a shot with a USB DVD drive and see if there are any differences versus using the installed (SATA/PATA) DVD drive.

I got the device ID using lspci. The vendor ID and device ID were already listed in ata_piix.xml, but I assume that because HP uses a sub-vendor and (I assume) sub-device ID such that the generic module wasn't getting properly matched to the HP-branded device.

I tried this hack with both ESX and ESXi, but neither seemed to work. I also tried using a Linux live CD to chroot into the service console environment and when I ran esxcfg-boot and it gave quite a bit more output. I can post that output when I get into work tomorrow.

I also tried replicating the XML used by some of the other drives where a <table> tag is used, but that didn't seem to make a difference.

My attempts to use a USB DVD drive to install didn't seem to make a difference either... so my task for tomorrow will be digging into the HP documentation to see if their BIOS allows disabling "legacy" IDE/SATA mode.

So the problem at hand may not have been directly related to the ICH9 support or hacking PCI config files. After reviewing the boot scripts and the kernel output after a failed boot, it seemed like the /proc/vmware directory wasn't getting populated (although it did seem that /proc was getting mounted) as I was getting errors about /proc/vmware/pci not existing among other errors. Just prior to that, there were a large number of APIC-related messages. On a whim I added "noapic" as a kernel parameter via GRUB and this time the init scripts did seem to work and the VMWare kernel seems to have started. I'm currently reinstalling and will see if it works out of the box if I add "noapic".

Well, a follow up where I get to put my tail between my legs. I just got a bunch of new servers and confused the model numbers. I was testing with an HP DL320 G5p, not a DL180 G5. I'm using the DL180 to run OpenFiler. The upside is that the DL320 G5p does work with the "noapic" option (at least for evaluation and prototyping).

When using the "noapic" option, I did not need to add the sub-vendor PCI ID to get ESX to boot properly.