Server of Doom – Musings on the hardware

One of the reasons I like ZFS is the reliability of storing data on it – everything stored is checksummed and validated each time it’s read from the file system.

Aether currently has a failed drive on the OS zpool, which I noticed a few days ago when I logged into the server to poke at the settings:

aether:~$ zpool status rpool1
 pool: rpool1
 state: DEGRADED
status: One or more devices are faulted in response to persistent errors.
 Sufficient replicas exist for the pool to continue functioning in a
 degraded state.
action: Replace the faulted device, or use 'zpool clear' to mark the device
 repaired.
 scan: scrub repaired 0 in 1h22m with 0 errors on Tue Oct 15 22:38:56 2013
config:
NAME             STATE    READ WRITE CKSUM
 rpool1          DEGRADED    0     0     0
   mirror-0      DEGRADED    0     0     0
     c2t0d0s0    ONLINE      0     0     0
     c2t2d0s2    FAULTED     1   120     0 too many errors
errors: No known data errors

The machine has been happily running with one dead drive, but it needs to be replaced in case the other drive decides to join in the fun! I run a scrub on the pool after I noticed the degraded drive as a just-in-case, but it passed 🙂

I was planning on buying another drive for the machine, but I couldn’t find any 250GB laptop drives available. My first thought was that I’ll buy a larger drive (500GB-750GB range) now and a similarly sized one when I’m buying the hardware for the new server. Then I got to thinking what I’d use 750GB for on an OS partition – the FreeBSD install is about 1GB at the moment with xorg and xfce4 installed. With the current plan of running Virtualbox to run something like XBMCbuntu for media playback and either Ubuntu Server or Arch Linux for the internet facing services. Even allowing 8GB each for those OS’s, the total space needed (including swap space) is under 20GB. So buying even a 500GB drive (smallest current-gen hardware) would leave the vast majority of the drive unused.

My current thoughts are that I could use a pair of 120GB SSD’s, then partition them to use ~32GB for the root pool, then use the rest of the capacity for the L2ARC and ZIL for the main storage pool. Which should lead to the machine being ludicrously fast as the OS would be running off a mirrored SSD pool (ZFS alternates read requests between the mirrored drives) and there’d be 80+ GB of SSD cache and ZFS Intent Log for the main array 😀

But I digress – the current dilemma is what drive to buy to replace the failed one in the current server!

It seems slightly wasteful to buy a new drive now, knowing that it’ll not be used in the new server. I do have a 750GB drive that I’m using as an external USB3 drive – I’m thinking of just swapping that one into the server and putting the money I would have spent on a replacement drive aside for the new server fund.

Tagged with: ,

Leave a Reply

Your email address will not be published. Required fields are marked *

*