PFS2 disk errors unable to be fixed by diskvalid

I recently had to restart my A1200 tower as it had run out of RAM (yes, all 128MB of it from having >20 browser tabs open and other programs) but the HD light was on at the time (it had frozen and I couldn't free up any more RAM for the processes to finish).

Now one of the partitions has errors which "diskvalid" fails to fix.

The report from diskvalid is:

---------
checking reserved tree
getting rootblock
checking disk Work3
checking .deldir
checking anode tree
checking anode indexblock 0
checking anode indexblock 1
checking anode indexblock 2
checking anode indexblock 3
anodeblock 807 blocknr 1902 has an incorrect id
indexblock 3: child failure
checking bitmap tree
checking bitmap indexblock 0
checking bitmap indexblock 1
checking bitmap indexblock 2
checking bitmap indexblock 3
checking bitmap indexblock 4
starting directory tree check
anodeblock 807 blocknr 1902 has an incorrect id
finished directory tree check
encountered 2 errors of which 0 were fixed

----------


I don't ever recall diskvalid failing to fix errors in the past (in the rare occasions I've needed to use it when similar block error requesters have come up after a reboot).

I am unable to write to the partition and it comes up with "disk full" and "write protected" requesters.

Now I'm unable to see a display of what files are on the partition. Luckily it's not my YAM emails partition and mainly was IBrowse cache and downloaded files.

I have also tried running diskvalid from "no startup-sequence" at boot, but it made no difference to the attempts to fix the problems.

If it makes any difference, the drive is approx 20 year old 20GB Seagate Baracuda SCSI-II and is partitioned into four 4GB partitions and a smaller partition for the remainder of the size of the drive. It has never missed a beat (apart from the occasional errors mentioned before, caused by similar means). Normally the Amiga stays on 24/7 except for the occasional reboot from running out of RAM (about once every 6 weeks or so) or when I go away for more than a few days.

Since the problem, the CPU load has been constantly on 100% except when I run diskvalid on that partition.


I know pfs3 exists, but not sure if it's backwards-compatible with pfs2.

I have tried emailing support from the contact details on the webpage but the emails bounce.

Does anyone have any other suggestions I can try to salvage the files before I reformat it?

[thomas]

You could download PFS3 and try the PFSDoctor on the partition. PFSDoctor can handle PFS2 partitions, too.

And yes, PFS3 is fully compatible with PFS2. You just need to replace the executable with HDToolbox in order to upgrade. Both PFS3 and PFS2 are even compatible wih AFS.

If PFSDoctor fails or if you fear to loose files you can try PFSSalv from Aminet. It copies rescuable files to another partition.

If the partition stays write protected, you will have difficulties formatting it because the format program will complain about the protection. You'll need to change it to FFS in HDToolbox, then format it, change it back to PFS and format it again.

[Bamiga2002]

Yes as Thomas suggested update to PFS3 through HDToolBox first. Then run PFSDoctor rebooting without startup-sequence. It works best that way. And PFSSalv2 is your last resort if that fails.

[thomas]

Originally Posted by Bamiga2002

Then run PFSDoctor [...] PFSSalv2 is your last resort if that fails.

Well, if there are valuable files on the partition I would always run a salvage tool first before I attempt to repair.

The repair program will attempt to make the partition readable at the cost of lost files. Those files which cannot be repaired will be deleted.

The salvage program will only read from the damaged partition so that you can have multiple tries with different programs or different options to rescue your files.

Thanks guys. All the DL links on PFS webpages I found were broken so for the benefit of others reading this, here's the link directly to all PFS-related files on aminet:

http://aminet.net/search?query=pfs

I'll let you know how I go.

edit: Just looking at the files found on PFS Salv2 and there appears to be files and folders missing in it's display that I can see when looking at the same paritition using Diskmaster (file utility).

When trying to copy files manually using Diskmaster, I am getting "wrong ablock id" error reqesters on some files. Presumably those files wont be able to be salvaged with PFS Salv2 - particularly if those files (and even entire directories) aren't coming up in the PFS Salv2 display.

I wonder if PFS is somehow validating the partition as earlier attempts I was unable to see much of the files/directories which I knew were there. Now I've done another reboot and the CPU load has reduced back to normal.

[thomas]

There is a vague chance that the developer of PFSSalv2 is somewhere around. You could try to ask him for support. ;-)

[r.cade]

20 year old 20G drive?! I don't think that's possible. We were back in the 20 MEG drive era in 1992.

Originally Posted by r.cade

20 year old 20G drive?! I don't think that's possible. We were back in the 20 MEG drive era in 1992.

Perhaps it's not quite 20y/o then - got to be over 10 y/o; I know it cost me a bomb (~$450 vaguely from memory)

Progress (if you can call it that) so far: I've fixed the partition with the problems, but I now have problems with another partition. While copying files from one to another partition, I started getting "LRU Update failed" (or similar). This other partition not even PFSDoctor can fix (even though it says it fixes the errors, I still can't use the partition).

In a last attempt I tried to format the "new" faulty partiton after changing it's file system to PFS3, but I still have the same problem (rootblock errors) even though there's no files on it. Format doesn't work as it says the drive is write protected. HDToolbox doesn't appear to have a format option either.

I tried PFSDoctor without startup-sequence and that didn't make much difference. I've kept log files on another partition but haven't read through them yet.

I was starting to think perhaps the drive is on it's way out, however the last time I rebooted I couldn't see workbench (although I know it loaded as the hard drive light flickered, indicating the startup-sequence is being carried out). I tried a few times but still no display. I doubt it's anything I've done to the drive partitions, and merely coincidence that I can't see anything on the screen. I'll remove the monitor switcher and connect the monitor directly to the gfx card to see if that makes any difference (I'm guessing it wont, but I need to follow the process of elimination).

This is my typical experience whenever something goes wrong with a computer - one problem after another

Some more progress...

I've now reformatted the "suspect" partition. I have also discovered the "original" 1GB IDE drive (which used to live inside the A1200 when it was still a desktop, as a replacement for the 40MB IDE drive - and is more likely to be closer to 20 y/o than the 20GB SCSI drive I mentioned in an earlier post) has read errors on a couple of different blocks. I've even identified the affected file (it's borderline 4MB). This partition is also PFS2 and it's also my boot partition. I've copied files from it onto the newly-formatted partition on the other drive, but aren't game enough yet to run PFSDoctor on it I'll have to make sure I use HDToolBox to make sure the "new" partition is bootable first.

It turned out that the main boot paritition on the 1GB IDE drive was the suspect one. As I was only given a block number in the error requester, I had no idea which drive or partition had the error. I've since copied everything from that defective boot partition onto a partition on the 20GB drive and made that partition bootable and have successfully been booting from it for the last few weeks. Now it's just a matter of identifying which programs are launched using the Dock bar from the original partition and updating them to start from the new partition before I format the defective partition to try and "block out" the problematic block(s) - which I presume the OS or drive should be able to do automatically (or is such a feature not around that long ago?).

I'll get there eventually...