Re: RAID5 ... N/Work with 300 GB SATA drives ... Likely Hardware Problem
From: Blondie (nobody_at_nowhere.org)
Date: 03/14/05
- Next message: Jeff Teel: "Security Event Log Question"
- Previous message: Phillip Windell: "Re: Gateway issues"
- In reply to: James B: "Re: RAID5 ... is there a disk size limitation?"
- Next in thread: Blondie: "Re: RAID5 ... Hardware problem"
- Messages sorted by: [ date ] [ thread ]
Date: Mon, 14 Mar 2005 16:42:14 -0500
I did check the Disks using PowerMax (the failing drives are Maxtor SATA
drives) 90 second test ... once a week ago ... they all passed then, but
last night there was a 'correctable' problem with one of the disks ... last
night one of them got a message 'run full read scan test' ... which I did,
and the test said there was an error (no specific info) ... writing the
correction may cause loss of data, proceed Y or N?
After completing the PowerMax tests with no outstanding errors I moved the
4x 300 GB Drives to a different server (was running Red Hat till 2 days ago)
and installed SBS2003 to test the Software RAID 5 with different hardware
... it worked, built a complete RAID5 array with the 300 GB Hard Disks in
about 4 hours.
Now I have moved those 300 GB Hard Drives back into the server where they
have never worked yet ... 'un allocated' them and started from scratch to
build a RAID 5 Array ... not complete yet, but it is still running properly
(this server is much slower) ... about 5 hours already.
What ever the real problem is, it is related to these Maxtor 300 GB SATA
disks and maybe their iteraction with the Adaptec 8110 (on Motherboard) SATA
controller ... they only seem to fail on one of the two servers, they did
work as a RAID5 Array built by RedHat V3 a few days ago.
"James B" <nospam@here.com> wrote in message
news:uHZ1aTDKFHA.2628@tk2msftngp13.phx.gbl...
> You know I would run the disk tools for whatever drive manufacturer your
> using just in case it's a hardware problem. Not the first time a drive has
> arrived in good enough shape for basic use but when ran thru it's paces it
> fails and resyncing is probably touching each cluster of the drive(s).
>
> "Blondie" <nobody@nowhere.org> wrote in message
> news:%23IuMiDCKFHA.572@tk2msftngp13.phx.gbl...
>> Ah ... Hardware performance, I can actually understand that :)
>>
>> I don't need high data rates, nor quick response times for the primary
>> application. I do need better than average MTBF ratings, and at least
> some
>> of these new SATA drives seem to have respectable MTBF (not sure about
>> the
>> duty cycle though) ratings ... and decent warranty coverage to back up
> those
>> ratings.
>>
>> However I am still not willing to 'bet' that any particular hardware
> device
>> will NOT fail ... I think I need to plan for 'routine' recovery from any
>> individual component failure ... and think that RAID 5 does a lot of this
>> for me.
>>
>> I thought you were going to tell me that using CPU cycles for RAID 5
>> implementation instead of a Hardware RAID controller was a bad idea ...
> I'm
>> not sure about that trade off yet.
>>
>> PS ...
>> The 'new' server (Volume size = 931.54 GB) downstairs is now at
> 'Resynching
>> = 38%' ... with no detected problems showing up yet ... this is about 90
>> minutes after it started 'Resynching'
>> The server (Volume size = 838.42 GB) that failed is 'stuck' at 78% (has
> been
>> for almost 2 days) ... the mouse pointer still works, but cannot peform
> any
>> other actions.
>>
>>
>>
>> "SuperGumby [SBS MVP]" <not@your.nellie> wrote in message
>> news:Ol5oK2BKFHA.2728@TK2MSFTNGP10.phx.gbl...
>> >I do not believe drives of this type are designed for 24*7 server
>> >operations. They are basically standard IDE drives with SATA interfaces.
>> >
>> > I sometimes refer to some SATA devices as 'SCSIlike drives'. An example
>> > being the WD 10K SATA Raptors. By SCSIlike I mean:
>> > Spin rates (and therefore throughput, of the drive) similar to SCSI.
>> > Low seek times.
>> > High MTBF figures (this being a major factor).
>> > Native Command Queueing (though I believe this is making it into
>> > IDElike
>> > SATAs also)
>> > A coincidence of this is that the drives will have capacities similar
>> > to
>> > SCSI. ie. instead of 40GB they will quote 36GB, instead of 80, 72/73.
>> >
>> > HECK, at least you're on the ball enough to be RAIDing them.
>> >
>> >> re:
>> >> If you were using this for anything but 'copies of backups' I'd
>> >> suggest
>> >> it was foolish.
>> >>
>> >> I'm not sure if this advice or not :)
>> >>
>> >> I do think there will be more than just backup files on this RAID 5
>> >> Volume, probably my own user data files at least.
>> >> Is there a performance issue with this configuration?
>> >> Or does it look like a poor cost / benefit decision?
>> >
>> >
>>
>>
>
>
- Next message: Jeff Teel: "Security Event Log Question"
- Previous message: Phillip Windell: "Re: Gateway issues"
- In reply to: James B: "Re: RAID5 ... is there a disk size limitation?"
- Next in thread: Blondie: "Re: RAID5 ... Hardware problem"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
