On Tuesday, 8 December 1998,
     HERBERT ROSENAU wrote to IVAN TODOROSKI about TCP/IP

 IT>>   The mouse freezes?!? Ouch! I don't think WathcDog can help you
 IT>> then...   My experience with OS/2 is that as long as the mouse
 IT>> works, you know   that at least a part of the system still works,
 IT>> especially background   programs like WatchDog, even if the PM
 IT>> hangs. But if the mouse   freezes, then either you pulled your
 IT>> mouse cable too hard :) or the   system is totally, completely
 IT>> and utterly dead. The mouse is one of   the last things to go...

 HR> No. You can see it if you have a node and your keyboard AND mouse
 HR> are without any reaction - an then comes an incoming call ->
 HR> you'll see your modems activity and - after you reboot or telnet -
 HR> you can look in your mailer and tosser log for the activity.

   No, I wasn't talking about the mouse being without reaction bu still
   moving... if you can move the mouse, than the basic parts of OS/2 are
   probably still working, and only the WPS (or in more severe cases the
   PM itself) has died and is not responding to mouse input. That's why
   text mode apps continue to run in the background, and that's why
   telnetd still allows you to log in and work on the machine, because
   is has nothing to do with PM. You're machine is still very much
   alive, just the user interface is blocked, and sometimes you can even
   recover from this.

  What I was reffering to above is a COMPLETELY DEAD mouse, when the
  mouse pointer is frozen and doesn't move on the screen! In that case
  either your mouse (or the COM port to which it is attached) isn't
  functioning, or your system hit the end of the road...

  In my experience, when the mouse goes, it's time for a cold reboot and
  biting your fingernails waiting for CHKDSK to say something weird...

  HPFS is extremely robust, so this rarely happens, and when it does,
  you can still salvage almost all of your information, since the vital
  file system structures aren't centralized like the FAT, they are
  dispersed in F-nodes across the disk, usually close to the files they
  represent. There is also a great deal of redundant information, for
  instance every directory entry has a pointer to the F-node of the file
  it represents, and every F-node has a pointer back to its parent dir.

  So if you happen to lose your root directory, or some other directory
  high in the hierarchy, the directories and files below it are not
  completely lost! All you have to do is scan the disk for F-nodes
  (F-nodes have a unique 32bit signature, thus are quite easy to find),
  and when you find one of the lost files, you immediately have a
  pointer to the parent dir (and thus to all the files in it), the
  parent dir has a pointer to its parent etc, so you can quite easily
  reconstruct the rest of the file system in case of failure of a part
  of the disk surface.

  I wonder if EXT2FS and NTFS are this robust...

  If you lose the beggining of the disk on a FAT system (both copies of
  the FAT are in the beggining), you won't even get a chance to atleast
  kiss your files goodbye... :)


                                                            - Ivan -

.!. Put on your seatbelt. I wanna try something ...!
--- Terminate 5.00/Pro [OS/2]
 þ TerMail/QWK þ