• Welcome to OS2World OLD-STATIC-BACKUP Forum.
 

News:

This is an old OS2World backup forum for reference only. IT IS READ ONLY!!!

If you need help with OS/2 - eComStation visit http://www.os2world.com/forum

Main Menu

eCS 1.2MR HPFS CHKDSK and other problems

Started by RobertM, 2008.05.08, 06:49:25

Previous topic - Next topic

RobertM

Hey all,

I have two machines running eCS 1.2MR and am having some interesting issues with them.

The common denominator is that they both seem to be on the HPFS partitions. Neither machine is in any way similar hardware wise (one is an Intel board with an 82845 chipset, and IDE drives, the other is I *THINK* a Microstar, running one SATA drive, but am not sure - it is at a client's). Both machines are running the DaniDrivers that came with eCS 1.2MR.

On the Intel Machine:
The machine "crashed" (power failure) and on reboot, started to lose files on Drive D (disk 0) and placed them in \Found0 etc directories. It also seems in reading or writing files on Drive C or D (disk 0) that some of the files are corrupt.

On normal (and successful) shutdown and restart after that, Drive D is not usually accessible until I do a chkdsk. A chkdsk /c works in this occassion oddly (guess it just marks the partition as clean - as should have been done on shutdown).

This machine has two IDE drives, and one SATA drive. The other (larger) IDE drive does not seem to exhibit this behavior, but has a JFS partition (spanned across the SATA drive).

As a side note, I am replacing the first drive on the Intel system (in the hopes it is just failing)... now, here is another interesting thing... in trying to archive the data (boot off the CD, xcopy to another drive), Drive C went through fine... Drive D gave me a "cannot read from the selected device" error (paraphrased)... I tried to run a chkdsk D: /F (no locked files or anything - booted from CD)... it gave me the normal chkdsk message:

The current hard disk drive is: D:
The type of file system for the disk is HPFS.
The HPFS file system program has been started.


And returned to the command prompt... same as the second machine does on Drive C... after that though, the xcopy is going through fine - even though it didnt seem to finish the chkdsk.




On the other machine: (all HPFS)
The machine has crashed due to power outages as well (since resolved with a UPS of crappy quality - so I expect this issue to recur again)... on restart, the machine goes through the motion of chkdsk'ing Drive C,  and doesnt actually complete - it does successfully chkdsk Drive D and E though (same physical disk). Drive C remains unusable until booted from the eCS CD and chkdsk'ing it from there... why it works there - and not during boot, I do not know.

So far, this machine does not seem to have lost any data - but in the event power goes out for more than 10 minutes at this location, the machine will "crash" - and then try to (and fail to) restart itself when power is restored.

The second system has a brand new drive - and consistently can be chkdsk'd from the CD properly - but not during boot. (the last few times, if it is crashed, I dont even try to reboot - I just pop in the CD, chkdsk, remove the CD and restart).

I have not enabled any special flags for the drivers on either machine.


Any ideas would be greatly appreciated...
Robert


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


Saijin_Naib

If I recall correctly, to :actually: have a real check-disk run on the drive, you must boot from the CD. Since the C: drive is being used to load up OS/2 and run the pre-boot check, the volume is mounted. This is why the other drives will complete. Or, thats how I understand it. Thats why I keep my RC4 CD handy. We really, really, REALLY need a pre-boot chkdsk routine like Windows so I don't have to perform CD juggling acts :\

RobertM

Quote from: Saijin_Naib on 2008.05.08, 06:53:01
If I recall correctly, to :actually: have a real check-disk run on the drive, you must boot from the CD. Since the C: drive is being used to load up OS/2 and run the pre-boot check, the volume is mounted. This is why the other drives will complete. Or, thats how I understand it. Thats why I keep my RC4 CD handy. We really, really, REALLY need a pre-boot chkdsk routine like Windows so I don't have to perform CD juggling acts :\

Yeah... the odd thing is, I dont run into that issue with any other machine... this piece of crap used to crash all the time and restart itself fine, even with stuff loading from 3 different partitions... it would just run it's chkdsk's and restart the machine and boot on the restart. And I dont even wanna mention how many times my foot hits the power switch (and the idiot who works in front of the machine I still hasnt havent moved it from in front of their foot).

This is the first (two) times I have seen this happen - and only on eCS so far... the other machines I use are all WSeB.

Weird...


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


Saijin_Naib

Maybe its just an eCS specific issue then? I only have experience with eCS here, and my eCS exhibits this behavior with both HPFS and JFS formatted boot-drives. hmmm, I wonder what has changed from WSeB to eCS?

RobertM

Quote from: Saijin_Naib on 2008.05.08, 07:03:44
Maybe its just an eCS specific issue then? I only have experience with eCS here, and my eCS exhibits this behavior with both HPFS and JFS formatted boot-drives. hmmm, I wonder what has changed from WSeB to eCS?

That is what I am beginning to think... I know I could put WSeB on the Intel machine... but I know I cant on the other one. My other WSeB machines are running the same versions of DaniS506 as the eCS machines... so I dont think that is it.

I wonder if something changed in chkdsk on eCS.... either that, or it might be a switch on a driver that the eCS installer put in (I use none on the WSeB DaniS506 machines)... will check on that... if you are correct (which I suspect you are) then I think it would only point to 3 things I can think of... (1) a difference in LVM, (2) a change in the chkdsk components, and/or (3) some switch enabling/disabling support of something in DaniS506 (that I am not using on the WSeB machines).

I'll check and see if I can figure something out...


Thanks,
Robert


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


warpcafe

Hi Robert,

...since you're talking about WSeB being installed - is it HPFS386 or the "usual" one, that you are running?
Sorry if you already told and I simply missed out on it...

Cheers,
Thomas
"It is not worth an intelligent man's time to be in the majority.
By definition, there are already enough people to do that"
- G.H. Hardy

RobertM

#6
Quote from: warpcafe on 2008.05.08, 09:57:09
Hi Robert,

...since you're talking about WSeB being installed - is it HPFS386 or the "usual" one, that you are running?
Sorry if you already told and I simply missed out on it...

Cheers,
Thomas

Hi Thomas...

The two machines I am having problems with are both HPFS as it came with eCS. No driver updates or changes from the default v1.2MR install.

The WSeB machines were running HPFS as originally came with WSeB CP2 PF and didnt exhibit this issue - but are now running HPFS386 v14.039 (if memory serves) and still not showing these issues. I brought that up in response to Saijin to try to say that I hadnt seen this issue until the eCS installs (ie: not on the WSeB machines here). Which also probably wasnt relevant - because the hardware is vastly different.

Sorry to have confused the issue. Was really tired after a very long day when I started this thread.




The only thing I can think of is a flag for the IDE drivers that should - or should not - be set. I think the Intel machine is/was just a failing drive. I've replaced the drive, xCopy'd the C partition back, restarted numerous times without needing to run a chkdsk on Drive D. I may power it off and see if it chkdsk's it properly later.

So... that leaves me with the client's machine. My suspicions are based off the fact that the CD (while using a very default set of parameters for DaniS506) will complete chkdsk properly - but it wont during boot. Those suspicions could be totally wrong though - and the machine isnt here for me to check what's in the config.sys file.

Fortunately, unless they have a big power failure, it doesnt matter (yet).

We've got 3 other boxes installed, same version of eCS, same or similar hard drives, same partitioning - but different motherboards in each... which dont show this problem. Two of them have been running for a couple years now, with only occassional reboots for database/support dll updates (worked our way up from MySQL 3.23.50/RxSQL - to v5 and the RxSQL that uses it).

I think what I need to do - unless someone else has had a similar problem - is make a trip out there and see how the IDE drivers are configured.

(I only posted without that info, because I thought the Intel machine here had a similar problem and I could have checked that here - but I am really thinking it was a bad drive).

Robert


Of course, you call can wait till I am more awake and try to re-state this in a fashion that actually makes sense...  ;D


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


Criguada

Quote from: Saijin_Naib on 2008.05.08, 06:53:01
If I recall correctly, to :actually: have a real check-disk run on the drive, you must boot from the CD. Since the C: drive is being used to load up OS/2 and run the pre-boot check, the volume is mounted. This is why the other drives will complete. Or, thats how I understand it. Thats why I keep my RC4 CD handy. We really, really, REALLY need a pre-boot chkdsk routine like Windows so I don't have to perform CD juggling acts :\

Saijin,
maybe I do not understand what you're saying, but it seems to me OS/2-eCS is not really lacking in this respect.

Provided you have the /AUTOCHECK parameter in your config.sys, OS/2 does a complete chkdisk upon boot if it finds that the file systems have the "dirty" flag (i.e. they haven't had a proper shutdown).
When talking about HPFS, IIRC you can even force one or more specific partitions to be checked on next boot even if they don't have the dirty flag set.

e.g. let's suppose you want to force chkdisk on partition "C" on the next boot; you only need to add a "+" in front of the "C" letter in the /AUTHOCHECK parameter on the IFS line:

IFS=HPFS /AUTOCHECK:+C

This can be done for JFS too, but is has a slightly different meaning, since JFS always does a quick consistency check on the log upon startup. With JFS, if you add a "+" in front of the partition letter, you're saying that you want to perform a full integrity check instead of the quick log consistency check.

Note I am writing this right out of my (faulty) memory... please check the docs.

OTOH, the problems described in this thread are of a different kind IMHO: in this case the chkdsk procedure isn't correctly terminating, and I can't really think of any plausible reason. I have never seen something like this.

Bye
Cris

Fahrvenugen

I ran into a similar situation on  a server that I maintain, only the partition that won't chkdsk isn't the boot drive (all HPFS partitions).  It would check the boot partition fine, but any other partition it would fail to properly check if it happened to  be booting, but would work fine if booted and chkdsk run from a CD.

What I ended up doing to work around it is, in my IFS line for the HPFS driver I put a specific /AUTOCHECK:C (so it would only attempt to check the C partition upon boot), and then in my config.sys file I put a:

call=c:\os2\chkdsk.com D: /c

After that, no more boot or chkdsk issues.

I don't know if that'll work for you or not, but it might be worth a try.


RobertM

Hi Fahrvenugen,

Sadly, the machine that has problems chkdsk'ing Drive D was a bad drive. The other machine, it is the boot drive that it fails on.

:(

Thanks,
Robert


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


Pete

Hi Robert

I noticed this thread the other day and can state that I have seen something like this before; as I could not remember any detail at the time I've had to have a think about it...

I suspect that you may find the disk that fails to complete chkdsk during boot has a partition table problem - I could be wrong but think you should run LVM and see if that reports any problem. If it does DFSee will be the next bit of software to run.

Alternatively, xcopy all files off the partition (while booted from another drive/CD) delete and then recreate the partition/volume, format and finally xcopy all files back.

Regards

Pete