[ale] Drive/Filesystem problem...

Jeff Hubbs Jhubbs at NIIT.com
Tue Nov 23 14:38:32 EST 1999


Is this single 40MB drive the only drive in the machine?  I'm not certain
you can install and boot to a partition that large (i.e., one that extends
past the 1024th cylinder).  You may need to install with a /boot partition
of about 19MB in size and then make at least one swap and one Linux native
partition as /.

Now, if you're trying to set this machine up for real, you probably should
have separate partitions for /usr, /var, /home, /etc, and perhaps even
/usr/local and /usr/src, probably broken up onto more than one drive.  Since
you care about testing bandwidth, if you are going to use IDE drives, don't
put more than one on a controller (i.e., don't go master-slave - I've been
looking into this very issue lately in conjunction with research into
software RAID).

I would go a little further and say that this single 40GB drive isn't a good
way to proceed even at this early a stage (again, presuming you want to test
bandwidth, not just functionality).  If you chop the drive up into separate
partitions, your heads are going to spend a fair bit of time slinging from
partition to partition.  It would behoove you to instead go SCSI and use a
handful of smaller drives, even if all you use is 4GB drives.  

Further still, if the point of your exercise is bandwidth assessment, then
using IDE drives in the first place probably isn't useful to you.  If it
were me and I wanted to go for max BW at low cost, I would either go for one
of the RAIDZONE subsystems (that use IDE drives - 200GB for under $7500
http://www.raidzone.com) or throw on a hardware RAID controller (Mylex?) and
set up a RAID 0 volume (if you can do that on a Mylex).  It will be
unreliable, but it will be fast.  I guess what I'm trying to say is that if
you are trying to demonstrate how fast a system can be in the disk i/o
department, you're not really doing an effective demonstration using a
single big IDE drive.  

- Jeff


-----Original Message-----
 From: Robert L. Harris [mailto:nomad at rnd-consulting.com]
Sent: Tuesday, November 23, 1999 11:19 AM
To: Atlanta Linux Enthusiasts
Subject: [ale] Drive/Filesystem problem...



Hey all,
  I have a friend setting up a dataware house and I convinced him to 
put it on a Linux box.  Right now it's not critical so we're just
testing bandwidth, etc.  At any point, he has a 40 Gig drive in
it and wants to keep it as a 40Gig disk, not a collection of 4s, etc...
He is having some problems.  Here is the output he sent me:


> EXT2-fs error (device ide1(22,1)): ext2_check_blocks_bitmap: Wrong free
> blocks count for group 0, stored = 31727, counted = 31740
> EXT2-fs error (device ide1(22,1)): ext2_check_blocks_bitmap: Wrong free
> blocks count for group 53, stored = 31740, counted = 9202
> EXT2-fs error (device ide1(22,1)): ext2_check_blocks_bitmap: Wrong free
> blocks count in super block, stored = 9689889, counted = 9667364
> EXT2-fs error (device ide1(22,1)): ext2_check_inodes_bitmap: Wrong free
> inodes count in group 0, stored = 32693, counted = 32704
> EXT2-fs error (device ide1(22,1)): ext2_check_inodes_bitmap: Wrong free
> inodes count in super block, stored = 10007413, counted = 10007424
> 
> Those are my main errors.  I get a lot of inode errors also.  And a ton of
> errors if I do an fsck.
> 
> The physical geometry of the drive is 16383,16,63, which is what ever
drive
> 18gig is set to.
> 
> The LBA geometry of the drive is 4982,255,63.  I calculated this myself.
> 
> The computer is a PII-233 not overclocked using a tekram P6L40-A4 or
> P6L40-A4X Motherboard.  I know there are bios upgrades for this
motherboard,
> but I don't know if they fix this problem.
> 
> The harddrive is a 40gig maxtor model 94098U8.
> 
> It partitions, and formats fine, but then after the format if I try to
mount
> it, it gives me errors like the one above.


Anyone have any ideas?  We need to hammer this out quickly or his boss will
start pushing for NT again.

---------------------------------------------------------------------------
Robert L. Harris                |   "A person is smart;
Senior System Engineer          |        People are dumb, panicky
  R&D Consulting.               \_            dangerous animals"  - Agent K



http://www.rnd-consulting.com/~nomad

DISCLAIMER:
      These are MY OPINIONS ALONE.  I speak for no-one else.

FYI:
 perl -e 'print $i=pack(c5,(41*2),sqrt(7056),(unpack(c,H)-2),oct(115),10);'






More information about the Ale mailing list