[NOTE TO READERS: First part talks about my experience with IBM drives and
about a tool for checking IBM drive media. Second part a little about XFS
and ReiserFS again, this time discussing on-disk format.]
On Mon, 30 Jul 2001 at 16:11, Juan Miguel Cacho wrote:
> In the price list of PCExpress they have an IBM 20.4GB @7200 RPM for
> P4900+ I was thinking of getting this drive, any comments would be
> apreciated.
I used an IBM 20GB 7200RPM UDMA/33 hard drive on my first Linux server,
and it ran happily for quite some time (I don't exactly know how long,
though). That's been decomissioned and is now a standard workstation. I've
transferred the hard drive to our design workstation and it's still doing
great. Gusi, my new server, has four IBM DTLA hard drives that have been
okay so far. One drive would keep going offline for awhile, because I
presume of bad block issues. Rebuilding the array (no data lost) fixed
things, and for a month now I've hadn't had any array failures. I presume
the drive marked the bad blocks and got some from its allocated spare.
My quirk with IBM drives is that local support seems to be fuzzy. IBM, for
example, will not replace the drives unless they came with an IBM box (ie:
Netfinity, et al). Unfortunately I don't have such a box. This I built
with parts from various sources.
IBM also provides a drive tester that can do pretty comprehensive drive
examinations. I have not tried getting this to work with the 3ware
controller, but for controllers where you can actually address an
individual drive (ie: most situations) then you don't have a problem. I
forget what this tool is called but I downloaded it from the Internet
(IBM's site of course) as an image of a boot floppy named dft-v210img.bin.
So I have v2.10 of dft. I forget what dft stands for. :)
> PS: Is it true that if you are using ReiserFS and your drive develops
> a bad block you are totally screwed?
Yes. I don't know what happens with XFS and JFS, though. Fortunately for
me the 3ware controller shields the rest of the system from such hardware
failure as this. Unless I have two drives offline at the same time, and as
long as the on-disk format of filesystem isn't buggy to begin with, and as
long as the filesystem code in the kernel doesn't decide to randomly write
garbage, I'm confident about the data in my system.
When any drive in my array hits a bad sector, that drive is marked as
offline and any access to that drive ceases. Fortunately the 3dm (3ware
disk monitor) web-based tool allows me to re-incorporate this drive into
the array and rebuild the array. I'm using RAID 5, so data on this
previously offline disk is rebuilt using the XORs of every bit in the
three other live disks. Yes, this takes quite awhile and hogs the system.
Fortunately I can configure 3dm to either prioritize disk I/O or rebuild
speed.
BTW, just for those who may want to know: one of the other reasons (aside
from the benchmark results and the statistics of files in my system that I
previously posted) why I prefer XFS over ReiserFS is the maturity of the
on-disk format. Unlike ReiserFS which has over the recent past already had
three official hashes and needs you (as far as I know) to remake your
filesystem to use a new hashing, XFS shares the on-disk format of its
mature IRIX counterpart. This allows XFS on Linux to benefit from the
rigorous testing that XFS on IRIX has undergone with all the movies SGI
boxes were used. While the filesystem code in the kernel is significantly
different and has recently changed much more than ReiserFS has (hence less
"stable"), the on-disk format has remained the same.
--> Jijo
--
Federico Sevilla III :: [EMAIL PROTECTED]
Network Administrator :: The Leather Collection, Inc.
_
Philippine Linux Users Group. Web site and archives at http://plug.linux.org.ph
To leave: send "unsubscribe" in the body to [EMAIL PROTECTED]
To subscribe to the Linux Newbies' List: send "subscribe" in the body to
[EMAIL PROTECTED]
