• 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

Fixing JFS problems

Started by ivan, 2009.04.17, 12:01:04

Previous topic - Next topic

ivan

Does anyone know a reliable way of fixing problems with a JFS volume?

I have a disk, formatted JFS, that has developed a bad sector.  It will not mount and chkdsk /f aborts at the disk size stage.

If necessary I will just reformat and loose the few files since the last backup.

Even though I use only OS/2 I would hope there is something for recovering JFS volumes out there especially since there is the much touted bootable JFS for eCs.

Radek

It depends where the bad sector is. Try mounting chkdsk and then chkdsk /F from a console run from your CD. This ensures that nothing has been started from the partition and that the partition is now "pure data". If the chkdsk fails again, then the bad sector seems to be in the control areas of the file system. Most likely, you will need to reformat.

ivan

Hi Radek,

Thanks for your suggestions, that was the first thing I tried.  I know exactly what sector, thanks to DFSEE, but it appears there is no way of getting anything from that volume unless it can be mounted.  At least with HPFS you stand a very good chance of recovering most, if not all, of what is on the volume.

I can't help wondering what is going to happen when some ones boot volume clags out and there are no tools to fix it.  Yes I know there is DFSEE and Jan has done a wonderful job on it... but.

Criguada


jep

Hi,

Yes, do try JRescuer, but if you want to recover files with filenames you have to regurarly run a script to set the filename (and path?) in the EAs of every file (Long Filename that is). JFS doesn't store such information by default. You can find a script with a progress bar in the Rexx forum/section unless you use the one included with JRescuer (setln.cmd).

//Jan-Erik

Quote from: Criguada on 2009.04.17, 16:21:31
Hi,
did you try with JRescuer?

http://en.ecomstation.ru/projects/jrescuer/

Bye
Cris

ivan

Thanks Criguada and Jep.  Yes I did try JRescuer.  All it produced was a long list of file names which when piped to a text file shows everything that should be there but no way could I get it to copy those files to anywhere else.

It looks as if I'll have to bite the bullet and reformat, at least it only two days worth of edits to re-do.  Then I'll remind myself that JFS is only useful for files larger than 2GB.

Saijin_Naib

... and boot times and file access speeds :P

ivan

Which is more important: file security or speed?

Criguada

Did you try to contact Jan (author of DFSee)?

ivan

Thanks everyone. 

Because this is on a production machine I formatted the drive restored from backup then moved everything, with the exception of two 3GB vdi files, over to an HPFS volume.  It turns out there were only two files that had to be updated and we had records of the edits.

Looking at the lack of tools to deal with problems on JFS volumes (work on JRescuer seems to have stopped and DFSEE is not geared up for it, though I may be doing Jan a disservice by saying that as he may have extra JFS problem solving in the works), I think the community is heading for problems (bootable JFS) but time will tell.



DougB

I think a little more history may be required, before you say that JFS may be a problem. It is known, that the older JFS did have some major problems, and what you are seeing, *may* be the result, if you did not reformat the volume, when the fix was made, after you did the update (this all happened a few years ago, so I don't remember the details). The "new " JFS (the bootable one), seems to be rock solid, for me (perhaps I have just been lucky). It can also be developed further, since they have the source, which is not true of the older IBM version.

On the other hand, you say that there was a bad sector. Depending on, exactly, where that sector is, and what was wrong with it, it may have taken out part of the stuff that is required to be able to recover anything. That is a hardware problem, and can happen to any file system, although a RAID system is supposed to be able to recover from that sort of thing. I do hope, that you replaced that disk, since one bad sector is a warning, that more may happen (may not too). At least you had a good backup, which is really the only recovery for hardware problems (and other things) on a disk drive.

ivan

Hi Doug,

You could be right, the machine is, with the exception of fixpacks, as it was created from ACP2.  As for the bad sector, a new disk is on order and the old one will be replaced as soon as it arrives.

QuoteAt least you had a good backup, which is really the only recovery for hardware problems (and other things) on a disk drive.

I couldn't agree more.  Here everything is backed up overnight every two days - maybe we'll do it every night when we get the new AMD quad core up and running to my satisfaction and I can assign a couple of 1TB drives to the task.

From what I read the bootable JFS is only available with eCs, which we don't have, and I'm a long way from being convinced to change over when all we might be interested in are the addons.  Now if they were available separately, that would be a different story ...
 

RobertM

Hi Ivan,

I had similar problems with eCS 1.2MR's JFS and the JFS that came with ACP2. The fix I found was to update the kernel and JFS using the update programs I mentioned elsewhere - manually updating them on the eCS machine caused nothing but crashes during boot.

I was able to recover virtually every file using JResuer once I purchased a license. My ACP2 JFS partition serves a few gigs a week (incoming and outgoing - via FTP), and has not exhibitted problems since I updated the kernel and JFS. I also limited the cache lazywrite settings on the commandline per someone else's suggestion to:

/CACHE:65536 /L:5,10,4 with the /L:5,10,4 part being the most important.

Not sure if that still applies to the latest JFS - but not up for taking a chance, as it took days to recover the eCS 1.2MR JFS partition - and I cannot afford even minutes of downtime on the ACP2 machine.


Rob


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


ivan

Hi Robert,

Nice to know I'm not alone in having problems :-)

Quote
I was able to recover virtually every file using JResuer once I purchased a license.
That may be the secret but nowhere in the JRescuer 20070208 package is there any documentation let alone any mention of it not working if you don't have a license. I have since downloaded the 20061213 package and there is likewise no indication that a license is necessary for it to work.  Ho hum.

The actual downtime on our MCP2 machine was zero! When they found they couldn't access the disk, I just switched off that disk caddy, pulled the disk and put in the spare, switched the caddy back on and everything was back to normal - less two days edits.
I don't know if this is normal but with this machine we've always been able to this and as it is on 24/7 i'm very thankful it does.
All my attempted recovery took place on a spare machine I'm testing, a quad core AMD with five 1TB disks, four of them in a raid array.  I might even consider a JFS volume on the system disk - if I find it stable enough. 
Thanks for the cache and lazywrite settings, I'll try and see.


RobertM

Hey Ivan,

I didn't see it in the docs either - did some digging to find out - after waiting for it to run through a whole list and do nothing.

Rob


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