Re: RAID on new 8-disk SBS server

Tech Tip: Click here to run a free scan for Windows Errors and optimize PC performance



I may have had two drives in a RAID 5 fail in a two-day period, about 3 years ago on an older PE server. What happened is that the first one failed, and the system rebuilt it on the hot spare. I replaced the failed drive and configured the new as hot spare. The next day, a second drive failed over to the new hot spare.

As soon as the second drive failed, Dell determined that the SCSI backplane was failing, and they replaced it. Either the backplane fried the drives, or the failing backplane falsely reported that the drives had failed. Because I had not lost any data, we never diagnosed further to determine whether the drives were actually bad. I got two new drives out of the deal, so I didn't care.

Anyway, Jason, the moral of the story is that IMO you can never go wrong with a hot spare. Even if you don't have multiple drive failures, it'll take some of the time pressure off replacing a failed drive.


"Leythos" <void@xxxxxxxxxxx> wrote in message news:MPG.224c012320e29c93989698@xxxxxxxxxxxxxxxxxxxx
In article <Ohp7yajiIHA.4080@xxxxxxxxxxxxxxxxxxxx>, "Larry Struckmeyer"
<lstruckmeyer(at)mis-wizards(dot)com> says...
Getting in late here, but RAID 5 without a hot spare is asking for trouble.
No one ever promised that two drives would not go out within seconds of each
other.

While RAID-5 has not been around as long as I have, in all my years of
doing computer work, since the 70's, I've never had two drives go out in
the same machine in the same week. The only time I've had a total RAID
failure was when a Noob connected 6 drives to channel A and 6 drives to
channel B on a controller, did a RAID-5 across both channels for 12
drives, and then had a bad cable about 11 months out - lost 6 drives at
once, but it was the cable not the drives.

When we don't have room for a hot spare, because of performance needs
for all slots to be active, we ALWAYS have a cold spare on the shelf,
and the servers always beep when a drive fails.

And there is also the false notion that RAID is somehow a backup method
:)

--
- Igitur qui desiderat pacem, praeparet bellum.
- Calling an illegal alien an "undocumented worker" is like calling a
drug dealer an "unlicensed pharmacist"
spam999free@xxxxxxxxxx (remove 999 for proper email address)

.



Relevant Pages

  • Re: HP proliant ML115 G5 on debian lenny
    ... dardrives to do RAID 1 backup for several other servers. ... simple backup server, any drive over 1 GB will hold all the ... system onto a 6 GB partition on one of the drives. ...
    (Debian-User)
  • Re: New Install of SBS 2003?
    ... What if I went with a raid 1 on two drives ... localy and have them install the sofware at my local computer guru's ... I do believe in overengineering ones' server. ...
    (microsoft.public.windows.server.sbs)
  • Re: Migrating to Raid
    ... >>> a RAID Mirroring environment. ... >>It is unclear whether you want to migrate the two existing 200GB drives into ... >uses FC4 as a file server, one uses SUSE as a web server, and one is ... >planning on upgrading to Win 2003 SBS from their existing NT4 setup. ...
    (alt.os.linux.suse)
  • Slow Performance on HP DL380 G4 SmartArray 6i Controller
    ... Hi all, I'm very new to SCSI and server hardware in general, so please ... as we only have the one SCSI RAID HBA in the system then I beleive the ... or the drives or cables attached to it. ...
    (comp.periphs.scsi)
  • Re: Need to migrate system drive from Raid 0 to Raid 1 - HP Smart Array 6400 ML-370 Ghost?
    ... I apologize - you are correct - I am looking to go from Raid 0 to ... I am actually using Windows Server 2003 Standard Edition - ... So you don't think any raid info from the 0 array would be on the ... other logical drives corrected to be Raid 0+1 for the mail store and logs ...
    (microsoft.public.windows.server.sbs)