[vox-tech] NAS/Printer Server/Web Server?

[Rick Moen]

Quoting Alex Mandel (tech_dev at wildintellect.com):

> I don't really know either, the only time I've had to mess with RAID was
> on a clunky old workstation with 3 drives and a terrible RAID BIOS.

Doesn't seem a big deal.

In my case, I'll need to muck about with /sbin/fdisk for a bit, because 
the mirrored pair has five filesystems plus swap:

linuxmafia:/etc/bind# fdisk -l /dev/sdc

Disk /dev/sdc: 18.3 GB, 18351959040 bytes
255 heads, 63 sectors/track, 2231 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x000c1659

   Device Boot      Start         End      Blocks   Id  System
/dev/sdc1               1        2231    17920476    5  Extended
/dev/sdc5               1         243     1951834+  fd  Linux raid autodetect
/dev/sdc6             244         425     1461883+  fd  Linux raid autodetect
/dev/sdc7             426         790     2931831   fd  Linux raid autodetect
/dev/sdc8             791         851      489951   82  Linux swap / Solaris
/dev/sdc9             852        1580     5855661   fd  Linux raid autodetect
/dev/sdc10           1581        2231     5229126   fd  Linux raid autodetect

I _could_ have made all of sdb and sdc be a single "Linux raid
autodetect" volume md0 and then fdisked that, but I didn't want to
mirror the swap space.

Given that setup decision, I'll need to power down, crack the case, yank
out and replace /dev/sdb with a fresh [sic] 18GB or so SCSI drive -- if
I still have any -- and make a partition table on it of the above specs.

Then, apparently I'll do:

raidhotadd /dev/md0 /dev/sdb5
raidhotadd /dev/md1 /dev/sdb6
raidhotadd /dev/md2 /dev/sdb7
mkswap /dev/sdb8
swapon -a
raidhotadd /dev/md3 /dev/sdb9
raidhotadd /dev/md5 /dev/sdb10

(Yeah, I know, 18GB SCSI drives in 2010 are a bit silly.  All I can say
is, time flies.  I should at least fish into the pile and see if I have
a pair of spare 73 GB ones -- though that will mean re-doing the
partition maps.)


> Single point of failure on the power supply doesn't worry me too much
> since I can live with downtime and it's seems to be a relatively
> ordinary part. 

No, you should worry.  Chief causes of HD catastrophic failure (other
than simple age and wear) are misbehaving PSUs and heat stress.

Putting both drives in a single external enclosure with a single PSU
means they can be both taken down at once by the same misbehaving power
supply or the same seized-up fan.  At that point, you have just
eradicated the entire point of having a RAID1 mirror pair.

> It will mean more work to configure since the NAS boxes seem to have
> print serve, file server, media serve all ready out of the box.

Yeah, well, for present purposes I'd rather run a Linux server using
simple, commodity, general-purpose components.  Works for Me.[tm]