Slow Imaging on ide disk

Question

I just changed form Zen7 Sp1 to Zen7 Sp1 HP2
I got plenty of Compaq D510 with ide drive.
With the sp1 only there were ok, but had problem with the DC7700.
Now when I'm booting with the d510 I got an error from udev
Udev event 3222 Create_node: symlink(../../hda,/dev/disk/by-path)failed file
exist
then i manually start a script that remove all partition an create 3
partitions on the drive
img pd-all
img pc1 ntfs 5000
img pc2 ntfs 5000
img pc3 ntfs
img pa1
when executing this script got many error in dmesg or f4 console
hda: task_in_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_in_intr: error=0x10 { SectorIdNotFound }, LBAsect=88539143,
sector=1162280967
ide: failed opcode was: unknown
hda: task_in_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_in_intr: error=0x10 { SectorIdNotFound }, LBAsect=88539143,
sector=1162280967
ide: failed opcode was: unknown
hda: task_in_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_in_intr: error=0x10 { SectorIdNotFound }, LBAsect=88539143,
sector=1162280967
ide: failed opcode was: unknown
ide0: reset: success
hda: task_in_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_in_intr: error=0x10 { SectorIdNotFound }, LBAsect=88539143,
sector=1162280967
ide: failed opcode was: unknown
end_request: I/O error, dev hda, sector 1162280967
hda: hda1
hda: hda1 hda2
hda: hda1 hda2 hda3
hda: hda1 hda2 hda3
But on the console there is no error, but after 2-3 minutes the prompt is
back and I can image.
On the old kernel take about 5-20 seconds.
When Imaging it take about 5-10 time longer then before I changed the
kernel.
was 5-6 minutes now takes 25-45 minutes
img restorep $PROXYADDR //$PROXYADDR/image/partition_c.zmg a1:p1
img restorep $PROXYADDR //$PROXYADDR/image/partition_d.zmg a1:p2
img restorep $PROXYADDR //$PROXYADDR/image/partition_e.zmg a1:p3
On sata drives there is no problem.
Is someone having the same behavior?
thanks

Answer 1

Pierre
We have managed to track down the issue. Basically, any partitioning
operation (img p...) resets the 'using_dma' flag
Using_dma is only relevant for ide drives and by default is set to 1 - this
increases img performance many times. As soon as we issue any partitioning
command it resets the value to 0, hence the poor performance.
We use ENGL ZIM in our environment and have modified our scripts to check
if using ide (i.e. /dev/hda as opposed to /dev/sda (sata/scsi), then reset
using_dma to 1 following the partitioning commands.
The command to check hdparm (assuming single ide drive - /dev/hda) is:
hdparm -d /dev/hda
To reset to 1:
hdparm -d1 /dev/hda
I shall report this to Novell and see what comes from it - I'm not sure why
it should be reset.
Hope this helps
Regards
David
> Pierre
>
> I have just heard, unofficially, that there are no existing bug reports for
> this issue, so it looks like we'll have to start....!
>
> I am back on the site that is experiencing this issue tom. and shall raise
> an SR. I'll keep you posted and see if we can make some headway on this
>
> Regards
>
> David
>
> "Pierre Boisvert" <[email protected]> wrote in message
> news:[email protected]...
> > No I did not raised an incident, but I will check to raise one. thanks, it
> > is good to know that i'm not alone...
> >
> >
> > "David Brightman" <[email protected]> a crit dans le
> > message de news: [email protected]...
> >> Pierre
> >>
> >> I can confirm this issue. We have ZDML7 SP1HP2 and have found the same.
> >> If you don't perform any 'img p' operations, and just let the base image
> >> pre-format the disk, etc. then it works at 'normal' speed. However, as
> >> soon as you perform any img p operations and then perform an img rp...
> >> a1:p1 it goes VERY slow - 4 times slower in our testing.
> >>
> >> The models we have issues with are Dell GX240, GX270 and Compaq D5xxx
> >> series (all ide). The SATA machines (Dell GX280, GX620, D520 and
> >> GX745's) are all OK
> >>
> >> Have you raised an incident with Novell? If so, I shall raise one and
> >> link to yours if you give me the details.
> >> We are implementing a new build process, with dual partitions and this
> >> is a big issue with us as there are many hundreds of machines with ide
> >> disks....
> >>
> >> Regards
> >>
> >> David
> >>
> >>
> >>
> >> "Pierre Boisvert" <[email protected]> wrote in message
> >> news:[email protected]...
> >>> Tested on other model today, and I got the problem on all PC with ide
> >>> disk.
> >>> When using
> >>> img pd1
> >>> img pd2
> >>> img pd3
> >>> img pd4
> >>> img restorep $PROXYADDR //$PROXYADDR/image/partition_c.zmg a1:p1
> >>>
> >>> got the error stated in the other post, but when I don't delete
> >>> partition and use the command
> >>>
> >>> img restorep $PROXYADDR //$PROXYADDR/image/partition_c.zmg
> >>> it is ok.
> >>>
> >>> I'M surely not alone in this case, does somebody else have this
> >>>
> >>
> >>
> >
> >
>