>  I *think* (I said "think") I have  found the culprit. The CMUCCVMA server was
>running  SP4 with  Rel.3 of  CARD MODULE  and a  clobbered A-disk,  which might
>account for the swapping of two blocks  of 400 bytes (although 400 bytes is not
>a  remarkable number  for CMS-format  disks --  unless the  disk was  formatted
>blksize 800, in which case it's a  half-track (see QQTRK &co) and it's a *very*
>reasonable  explanation)....
I believe (not 100% sure) that hashing doesn't apply to non-EDF format
disks. In any case, after I mentioned this CARD possibility to Eric
I discussed it a bunch w/ Larry Chace here at Cornell (who has been
quite deep into the new hashing code in dianosing a bunch of problems
with it) and we came to the conclusion that there was no way CARD
could be at the heart of such a swap. The sort of errors it would
produce would be much nastier rather that simple swapping of data.
In any case, it is certainly good to get that bad copy of CARD cleaned
up as the problems it can cause are VERY nasty (entire disks can be
lost before it becomes obvious that a problem exists).
 
ANYONE who has an old copy (especially if you are already at CMS Rel 4,
but even if you are at CMS Rel 3 so you don't have to worry about it
later) really should convert to the new version. You can either use
the copy that Eric has sent out or, if you want source and a HELP
file (perhaps Eric is already sending these?), contact me and I will
send you copies. -njg