kevin hamann

I have an approxmately 113 GB JFS drive. The computer had a trap 003
on Saturday. I stopped JFS chkdsk (autocheck as part of the boot
process) using cntl-alt-del on Sunday early (after about 12 hours of
jfs chkdsk).

I upgraded to the lastest versions of jfs.ifs, and ujfs.dll from a cd
boot. I then added BASEDEV=CHKDSK.SYS to config.sys and rebooted the
machine and let the autocheck process start. It took about 2 hours to
get through phase one and two. It's been on phase 3 for approximately
32 hours. How long does it normally take? I hate to have let it go
this long and then find out I need to start again.

Any and all replies deeply appreciated.

Kevin

marty

Sounds like you're in a serious world of hurt. Any idea if you're
having physical defect problems with your hard drive media? Is it
making any strange or disconcerting noises? I've got a mostly full 80GB
JFS partition here that takes about 30 seconds to CHKDSK.

The version of JFS that I've been using for almost a year without any
problems has been the 10/09/2002 version (packaged on 10/26/2002 IIRC).

paul ratcliffe

You have something major wrong. It should take seconds.

marcel_müller

Hi,


It mainly depends on the number of files and the free memory.

With around 1,000,000 files it takes some minutes on a Cheetah/15K here.

Without the option EARLYMEMINIT=TRUE in your config.sys the memory ist
limited to approx 15MB during boot. Chkdsk on JFS does not always handle
memory problems properly.

Since you cannot boot from JFS it is the best choice to disable the
autocheck for JFS generally. Instead, you can place something like
IF NOT EXISTS "J:\wproot. sf" THEN start chkdsk j: /f /c
for each jfs volume in you startup.cmd.

Note: chkdsk /c does the clean-check before execution even without the
IF EXISTS ... - However, the WPS does really bad things if it tries to
access a locked drive which was previously unlocked during startup. And
chkdsk /c always causes a short lock.

--
Marcel Müller

kevin hamann

I don't think the drive has any problems. 120 GB Western Digital with
the
8mb cache that I got about 3 months ago. No funny noises.

The JFS drive was made with the 5-8-01 version. The replacement
version
was from the eCS 1.1 cd (not sure on the date).

I rebooted the machine after 52 hours and then booted from cdrom and
am
trying the chkdsk from the cd. It took much less time to get through
phases 0-2 this
time. Hopefully it will finish overnight. I think, if nothing else
works, I'll get a program
to set the dirty bit clean and recover what files I want .

Thanks,

Kevin

kevin hamann

29,031,455 files on this 7.2k WD drive. 128 MB ram in the machine, so
probably
64 MB or so free.


That may very well have been part of my problem.

I think I'll set this up after the chkdsk I have running now finishes.
Apparently the
autocheck wasn't accessing the drive after phase 3 started ( I let it
run 52 hours
before quitting). I'm now running chkdsk from the eCS 1.1 cd and it
is accesing
the drive during phase 3 (I'm using the "turbo" led on the old AT
case in which
this drive resides just for the HD, the regular red hard drive led is
used for the
cd drives). It took less than an hour for phases 0-2 to run, and with
the drive
access during phase 3, I think I have a chance of it completing.

Thanks for all the information.

Kevin


--

mark dodel

-> On Wed, 17 Sep 2003 02:59:34 UTC, Marty <mamodeo@stny.rr.com> wrote:
->
-> > Sounds like you're in a serious world of hurt. Any idea if you're
-> > having physical defect problems with your hard drive media? Is it
-> > making any strange or disconcerting noises? I've got a mostly full 80GB
-> > JFS partition here that takes about 30 seconds to CHKDSK.
-> >
-> I don't think the drive has any problems. 120 GB Western Digital with
-> the
-> 8mb cache that I got about 3 months ago. No funny noises.
->
-> > The version of JFS that I've been using for almost a year without any
-> > problems has been the 10/09/2002 version (packaged on 10/26/2002 IIRC).
-> >
-> The JFS drive was made with the 5-8-01 version. The replacement
-> version
-> was from the eCS 1.1 cd (not sure on the date).
->
-> I rebooted the machine after 52 hours and then booted from cdrom and
-> am
-> trying the chkdsk from the cd. It took much less time to get through
-> phases 0-2 this
-> time. Hopefully it will finish overnight. I think, if nothing else
-> works, I'll get a program
-> to set the dirty bit clean and recover what files I want .
->

Since you didn't create the JFS volume with an early WSeB version,
this might not apply, but in the first couple of years with JFS it was
mandatory that CHKDSK HAD to be done from the same release that the
volume was created, or you would get the endless CHKDSK problem. By
"chkdsk from the CD" I assume you mean the MCP or eCS CD that was used
to create the volume. Hopefully that resurrects your drive.
Supposedly this was no longer a problem with the more recent releases
of JFS.

Mark

--
From the eComStation of Mark Dodel

http://www.os2voice.org
Warpstock 2003, San Francisco, October 18-19th -
http://www.warpstock.org

marty

I recall hearing of some funny business when moving from the base JFS
version included with eCS 1.0/MCP1 to a newer release, where it was
recommended that you reformat the JFS partitions with the newer version.
Just vague memories at this point so I might be mistaken.

francois fornier

Would you be running SCSI on an adaptec board for that drive by any
chance? If yes, what version of the driver are you running?

I have seen very slow disk checks on one machine before (MCP1 times). I
took several minutes to check a 1.5GB JFS partition! (9GB Fujitsu HDD
connected to an OEM AHA29160 SCSI adapter).

HTH,

--
be seeing you,
Francois - Francois dot Fornier at gmx dot net
(add punctuation to reply)

kevin hamann

I read similar things that's why I upgraded the jfs.ifs and ujfs.dll
when I created the
partition with the 5-8-01 version (not the version that came with eCS
1.0). I first
tried running chkdsk with the 5-8-01 version (got stuck on phase
three with no visible
disk access)and then tried it with the eCS 1.1 version put on the hard
disk (which is
considerably newer). It also got stuck on phase 3 with no visible
disk access. I then
booted from the eCS 1.1 CD and ran chkdsk from it. It's been running
for 27 hours and
is still on phase 3, but the led that corresponds to the hard drive is
at least blinking, so I'll
let it go til the end of the weekend I think.


If it still hasn't finished I think I'll try resetting the dirty bit
and see if I can access the
drive that way. I put together this particular machine to try jfs, so
If I lose the partition,
I haven't lost anything terribly important luckily.