Strange issue with PFS3-AIO
 

I just had this problem happening and it's really weird, perhaps some light can be shed here!

I had a corruption issue with my CF card and re-did it. As far as I could tell, the device works fine, because I was able to image it and fully format it without errors.

Yet I was just copying some stuff over, and these warnings appeared (in this order):

http://i.imgur.com/MQQxXa7.png
http://i.imgur.com/QDiA8gj.png
http://i.imgur.com/DZaj5Z6.png
http://i.imgur.com/TiItjH7.png
http://i.imgur.com/5cYqIUL.png


After this happens, the volume gets write-protected until I reset.

Weird thing is, this only happens when I want to copy a particular folder's data onto this CF partition. I can keep copying other stuff after this. It has not complained.

I am using WinUAE for this, 3.5.0, the system is an emulated A600 with KS and WB 3.1, 020 CPU on fastest speed, CF mounted as "a600/1200 IDE" and CF.
Sandisk 1GB card, set up with 10 buffers on partition.
I am copying from a PC directory mounted as a drive.
I will now try to copy that same folder to another location on the partition and see what happens.


[edit] I copied the same folder to the root of the partition, then moved it to where I wanted to put it originally and it worked. So I am a bit puzzled.

Obvious, isn't it?


RTFM!

Documentation probably is including the worst case, very tiny buffer works fine in most normal situations.

I guess buffer automatic increase (if memory is available) would be the best option. I'll think about it..

Why the rudeness? It isn't obvious. (and that documentation does not come with the PFS3-AIO archive, so I never read it, I don't use PFS3 v5.3) We discussed this in the past, this is why I chose 10 buffers. Normally I would have left it at the 30 default, and if I had shit tons of RAM to spare, I would have upped it. But I don't. I don't want to use 300KB of RAM for buffers on each partition.

So if the issue is created by buffers, I'll increase them. But I don't know, was the 30=150 issue ever fixed/resolved? Maybe I can spare 100- buffers. Gotta test.

I realize this is related to directories with loads of files, and I can curb that. But I never had all the other issues before when I ran out of buffers. The directory structure got corrupted before PFS3 could lock the drive. I don't even know what the other stuff meant, but I experienced some more issues later (ablock id issues), and I used the "diskvalid" tool to fix it. It now seems to be working fine, but I guess it could turn to issues down the line.

Did you try checking the partition with pfsdoctor?

As I just wrote, I used diskvalid and it fixed two errors, now everything is working well.
I'm trying to understand the situation so I can take measures to prevent it from happening again. I'm increasing buffers right now to see what's up.

Well, scratch my previous message.
I started getting a "wrong blockID" error again on another folder, and diskvalid would find errors but not fix them.

So I ran PFSDoctor, and it found a shit ton of errors (mostly "reserved block not used but allocated" and "anode not used but allocated"), and "fixed" " them, which is PFSDoctor for "I erased most of the shit in your hard drive, buddy" (all the WHDLoad games :mad) so now I gotta do all that again.

Before I take on a fool's errand, can anybody tell me if everything looks OK with my drive setup? I attached the RDB of the partition. I wonder if the maxtransfer value is wrong. I already changed the buffers to 100.

Hi AK ,which program do you use to view the RDB ?

Thomas' TSGUI: http://thomas-rapp.homepage.t-online.de/download.html

Thing is, I can't see it, this is why I posted it, for Thomas or someone else to dissect. I can't make head or tails of it. It's just a bunch of binary gibberish to me :lol :sad

Bump because I got no answers ☹

You said you would test a different setting for buffers. Can you report the results?

I guess out of buffers situation is something that shouldn't happen and when it happens in wrong place and time, it can mess up the filesystem metadata, for example if some update was aborted due to buffers running out because number of buffers was too small for worst case scenario.

Yeah, I said I had already raised the buffers to 100 and the result is post #7
This is why I posted my RDB for someone like Toni or Thomas to have a look.

Toni: could the system had gotten ruined from the get go and now will experience problems all the time?
I just don't want to copy stuff to the drive again to run into this issue one more time, it's a huge waste of time. I wanna do it once more and I would like to know if I should be safe copying files over again to replace all PFSDoctor erased, or if I should start from scratch and format and install again that partition (system partition seems OK, thankfully)

Bump? anybody had a look at the RDB?

This is interesting to me because I have had something similar happen most recently (just prior to disassembling my Amiga).

I was worried it's an image corruption.

for me, the situation is this:

A500 with V500v2+ using CFCard/SD adaptor on the IDE controller of the Vampire.

I got a wrong block ID error and ran PFSDoc. A ton of errors, same as you posted (Allocated but not used) and I think all of them were in WHDLoad/Games drawer.

As soon as my Amiga is back together I was going to install a fresh image of ApolloOS r43 (the one I am currently using that gave me errors). Once I put the fresh image on the system I will immediately run PFSDoc before any errors appear to see if it's a base image issue or something else. I haven't checked my buffers. Will have to wait until my A500 is up and running again.

Very interested in your outcome.

just one thing : never use a tool like disksalv or Quaterback on a PFS3-AIO partition.

This will corrupt the sectors .....

This has nothing to do with this thread.

well, yes, but it's always good to remind people this :)

However, this issue is strange.... What kind of hardware do you use to be able to plug an SSD disk in your Amiga ?

What was the outcome in your case? I'm getting the same problem from my ApolloOS installation, also from WHDLoad/Games directory

This is a known bug with the ApolloOS r43 image, so someone made a fixed version.