Dell R610 Debian Lenny issues
What can I say... what a pain.
I've installed Debian, Gentoo and RHEL on many servers, but this one Dell box caused so many issues:
- Firstly, the dell machine had 4Gb RAM originally and had a further 8Gb added. This was ok but guessing the configuration using trial and error was time consuming. And before you say RTFM I would draw your attention to the "insert memory in sequential order starting furthest away from processors" - bollocks - well ok, you need to factor in different memory sizes putting matching sized sticks in parallel configurations for dual processor machines.
- The internal SAS controller needed replacing with the new PERC 6/i controller. This too was ok but obviously meant we'd need to re-install the O/S with the HDD running from the new controller.
So, here came the problem. The latest Debian O/S and the Dell R610 box seemed completely incompatible:
- the keyboard would lock up just as the debian-installer showed the language chooser
- when you could get further, the system failed to locate:
- any of the four broadcom network cards
- any of the harddrives as RAID
So, some Google digging found many people with similar but not identical issues. The most obvious was a BIOS update which proved to be a complete ball ache as without an operating system you’re shafted. I had to get a USB pen drive (you don’t get floppy disk drives on servers anymore) and download a HP based program formaking bootable USB drives that we could put the BIOS image on.
This BIOS update went ok, system started and allowed the debian-installer to start. Immediately the system detected the missing Broadcom NIC’s stating I needed to obtain the .deb based driver file bnx2-09-4.0.5.fw. I downloaded this on a Windows box and put this on the USB stick too and went back to the Debian install to complete the driver installation.
Soon after the debian-installer did a disk detection process and found the existing (to be deleted) RAID configuration. I realised at this point that it was missing the newly installed SSD drive which was going to be the primary bootable device. So, back to the controller BIOS to configure both a new RAID 1 on the existing SAS disks, and a single virtual disk for the SSD. Reboot machine, start installer, choose drivers and detect disks. Now two disks are found (or three if you still have the USB drive there, and four if you have the DRAC virtual image enabled) and you can pick which to install the O/S on. BTW installing on the SSD took a matter of minutes.
Almost finally, after the O/S was installed successfully, on reboot the system would lock up immediately after the DRAC card! It turned out that the controller BIOS has a “bootable device” switch on one of the management pages. Flicking this to the SSD made the system get a little further. It appeared that the GRUB loader was now operational but not choosing the right drive to boot from.
Further messing (trial and error mainly) made the system boot a little further but it appeared that the automatic GRUB installer chose some random drive locations to boot from. To make it worse it had also done this within the fstab file so it meant that all these needed changing too.
Finally the system booted – all I can say is updating the BIOS solved the most issues, everything else was just normal server and O/S ball ache-ery!
UPDATE
For lenny and other Dell servers such as the R300 series (inc R310) the updated bnx2 driver can be found here.