Debian on Dell PowerEdge R200

Posted by Jamie Neil on 17 February 2008

We're retiring some of the older equipment in our rack at the moment, and replacing it with a couple of shiny new Dell PowerEdge R200 servers. Being fairly new there isn't that much information out there on these boxes, and even less about running Debian on them, so I thought I'd post my impressions and results.

Our boxes have the following spec:

  • Dual Core Intel Xeon 3065 (2.33GHz / 4MB Cache /1333MHz FSB)
  • 4GB 800MHz ECC Memory
  • 2 * 250GB WD RE HDD (7200rpm / SATA2 / 16MB Cache)

The drives are from Dell, but the memory is third party (Transcend) as Dell still seem to charge a small fortune for 800MHz ECC RAM. Chassis seems to be very similar to the SC1425 boxes (which make up the majority of our servers), but with a few little improvements like snap fit drive cages and bigger cpu fans.

My first attempt at installing Debian Etch (4.0r2) failed pretty quickly when the installer couldn't see the DVD drive - it booted from it fine, it just couldn't find a driver. This turned out to be a lack of support in the standard Etch kernel for the new Intel ICH9 chipset. I thought about installing Lenny instead, but as these boxes were for production use it wasn't something I really wanted to risk yet. So I had a look around to see if anyone had an Etch installer patched with a newer kernel, and as luck would have it Kenshi Muto had exactly what I was after: Debian Etch d-i images with 2.6.24 kernel, designed to support the hardware in the Asus Eee. Install was very straightforward, booting of USB flash instead of DVD this time (although I expect DVD would have worked fine), hard disks, DVD and NICs all detected without problems. I partitioned the disks into two RAID1 sets: one for /boot and the other as an LVM volume for the other partitions. Once installed I tried downgrading the kernel to 2.6.22 from backports, and this appears to work just as well as 2.6.24, so hopefully there won't be any problems moving to Lenny when it's released.

Haven't done any real stress testing yet, but everything seems pretty stable with no unexplained crashes or odd syslog messages. Benchmark disk performance is not quite as good as I hoped, here's the results from bonnie++:

Version  1.03       ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
ganymede 6576M 36955 69 36940 8 24299 4 56920 94 57667 6 370.8 1
------Sequential Create------ --------Random Create--------
-Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
16 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++
ganymede,6576M,36955,69,36940,8,24299,4,56920,94,57667,6,370.8,1,16,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++

Hopefully real world performance will be better. The second box will have different drives, and maybe a PCI-E controller card, so I'll have to see how much difference that makes.

Comments

No one has commented on this page yet.

Post your comment