At 14:53 14-5-2003 +0200, Olaf =?iso-8859-2?Q?Fr=B1czyk?= wrote:
I have linux 2.4.20 + XFS from CVS 09-Jan-2003.
System is based on RedHat 8.0.
Motherboard ASUS A7V266A (via KT266A chipset)
2 IBM SCSI disks connected to Adaptec 2940U2W
CD-RW, DVD-RAM connected to the same adaptec card
2 IBM IDE disks connected to motherboard controller
If I copy disk /dev/hda to /dev/hdc:
cat /dev/hda > /dev/hdc the system becomes __highly__ unresponsive
May I suggest the following procedure for copying disks and information...
partition the second disk similar to your first.
Mount these partition under something like /mnt/new
And the other partitions underneath that... eg.
mount /dev/hdc1 /mnt/new
mount /dev/hdc5 /mnt/new/usr
After completing this procedure you can transfer the data.
(cd /;tar -cf - .)|(cd /mnt/new;tar -xf -)
This will give you all files on the second disk. However you must reinstall
lilo on this disk since it will not be in the same place anymore.
Boot your favorite rescue disk (the installer in rescue works as well)
Asuming the XFS installer disk (skip partition check).
Mount the rootfs and reinstall lilo (after switching disks ofcourse)
mount /dev/hda1 /mnt/new
chroot /mnt/new /sbin/lilo
That should reinstall lilo for you.
Processor is about 90 % idle
Load average is about 5
Have you also checked the use of DMA on the IDE disks?
You can use hdparm /dev/hda to check for it.
Without DMA the IO susbsystem will be very slow on IDE systems.
hdparm -d1 /dev/hda
hdparm -d1 /dev/hdc
Should activate it. However note that DMA might be disabled by the kernel
because the disks and controller might be in the blacklist.
The VIA IDE driver has a scarred history.
Does anyone has idea why is it happening? Or maybe this is "normal"
Not it should not be enough to make it that severe. I suspect the dma issue.
I will have some free time later and I'll grab latest CVS and try it.
That is a good idea as well.
It might just be your lucky day, if you only knew.