Submission for Unix-PC

[gatelist at zorch.SF-Bay.ORG]

> Regarding BAD FREE LIST errors
Problem simplification!

> An fsck -n generated:
>   ...
>   ** Phase 4 - Check Reference Counts
>   UNREF FILE I=2873  OWNER=root MODE=10000
>   SIZE=0 MTIME=Dec 12 03:19 1989 
>   CLEAR?  no
> 
>   FREE INODE COUNT WRONG IN SUPERBLK
>   FIX?  no
>   ...

> The MTIME of the UNREF file is precisely the time at which I was doing
> the fsck!  I have not seen that kind of error before tonight.

Further analysis shows the Phase 4 error to be transient; it only occurs
when I am piping fsck output to a tee (and therefore writing to the tee
argument).  It does not occur at the simple "fsck -n" or "fsck -n >
fsck.log".
A better fsck output (than the one I posted earlier) to examine is:

  /dev/fp002   (NO WRITE)
  File System:  Volume: 

  ** Phase 1 - Check Blocks and Sizes
  POSSIBLE FILE SIZE ERROR I=278

  POSSIBLE FILE SIZE ERROR I=3076

  POSSIBLE FILE SIZE ERROR I=3097

  ** Phase 2 - Check Pathnames
  ** Phase 3 - Check Connectivity
  ** Phase 4 - Check Reference Counts
  ** Phase 5 - Check Free List 
  1 BLK(S) MISSING
  BAD FREE LIST
  SALVAGE?  no

  3253 files 51550 blocks 19162 free

Again, I will be grateful for enlightenment.

John R Ruckstuhl, Jr, ...!hplabs!hp-lsd!sphere!ruck