[SATLUG] RAID5 Recovery - Reassembly questions
adlabens at swbell.net
Wed Aug 26 18:42:15 CDT 2009
You ask some very good questions, & I'll try to answer them as best as possible.
RAID, by definition is supposed to be inexpensive disks. We had about
250 gb data, at the time I built it, and I wanted something bigger, but
didn't want to spend the immediate $ on tb drives, so I got 250 gb
drives and RAIDed them. It was less expensive at the outset. As many
drives as I've had outlive their technological lifespan, I never
figured a WD drive would fail in the first year, so I figured I'd have
time enough to buy 1 tb drives for backup at a cheaper price as the
year went along. Silly me! I also figured that as I purchased 1 tb
drives to use as backup, and filled up the 750 gb that the 4 250 gb
drives would create, I'd be able to eventually use what would be the 4
1tb drives I'd buy for backup as RAIDed drives to create a 3 tb RAID
array, and that would give the technology time to get more inexpensive
so that multi-tb drives would become more affordable. That was the
plan, and it seemed like a good one. To my mind, it still does. At
the time, 4 drives of 250 gb each were a total cost less than a single
1 tb drive. That was in November, so prices are dropping fairly
quickly. Today a 1 tb drives is about the cost of 2 250 gb drives.
However, as you've asked about the notifications, it was the
implementation that failed. I was a bit overwhelmed in building this
and failed to create the proper notification regarding disk failure.
HAD I done that, none of this would have been necessary to this
extreme. Lesson learned. It is, after all, a home server & not an
enterprise server, tho we use it as such. As for the notifications,
once this thing is back up & running, my immediate next orders of
business will be to setup and test the notifications, start doing
regular backups with a rotating group of 1 tb hdds, and install the APC
UPS (& the correct auto-shutdown routine) that I purchased about 2
months ago but have not yet installed (other than to plug it in and
have it connected to the house power).
Once I have all that done, I'll feel secure in knowing that nothing bad
will happen without me being notified and having a backup both at home
Had I felt that two WD drives would fail within the first year of
purchase & installation, I definitely would have moved forward MUCH
quicker with the backups & notification and the UPS & shutdown
routines. But, after having all my HDDs outlive the supporting
technology (by that, I mean, the OS's got replaced, requiring
motherboard upgrades and new HDDs to support them), I did not figure to
be due for a failure so quickly. In the true essence of the
philosophy: "Expect the Unexpected!" would have been the best thing to
do. I failed.
Would I do it again? Probably, but with a stronger desire to complete the ENTIRE project both correctly and more quickly.
San Antonio, TX
--- On Wed, 8/26/09, Herb Cee <hc at lookcee.com> wrote:
From: Herb Cee <hc at lookcee.com>
Subject: Re: [SATLUG] RAID5 Recovery - Reassembly questions
To: "The San Antonio Linux User's Group Mailing List" <satlug at satlug.org>
Date: Wednesday, August 26, 2009, 5:56 PM
David Labens wrote:
> So, to restore it (hopefully not needed), I'll simply need to put it back onto a drive of the same size (& all other parameters). It won't necessarily need to be on the exact same drive, will it?
> I'm continuing, working on moving drive 3 to partition /3 of the Big Disk.
David,Whew, following this has been very interesting but now .. I am even more confused about why use RAID in a personal puter network?
It just seems to me redundent since you are gonna have to use a backup system anyway to feel secure. I still do not have my little whirlygig working the way I want, but it really seems that using RAID would be risky for me since I would be so far over my head in attempting to administer the sys. On the other hand either setting up a frequent synced backup is safely within my understanding and tech ability.
The Raid sys also means keeping all the spindles spinning plus the access time for all heads to R/W so the overhead is higher. To my simple logic it appears that unless you operate a mission critical operation RAID is a potential drag.
After going tru this really long ordeal how do you personally think about RAID's value? Another Question is why did the sys not remind you about that early drive being dropped and not attended. Or was that just a failure to set the procedure to notify you?
SATLUG mailing list
SATLUG at satlug.org
http://alamo.satlug.org/mailman/listinfo/satlug to manage/unsubscribe
Powered by Rackspace (www.rackspace.com)
More information about the SATLUG