It has happened again on the ROOTS-L list.  This time, there was only
two hours delay between the time the message index went out and the
time the request went back.  The sequence numbers were off by one, as
if one message had been knocked out of the archive, but, interestingly
enough, there was a message in that batch containing a line of "====="
followed by a line beginning "Date:" (though not containing a date
formatted according to the rulsed of RFC822, by any means), so the
numbers got resynchronized by accident just before the end of that index.
For the record, here is time tag on the index mailing:
 
           Fri, 19 Nov 1993 20:18:33 -0600
 
Incidentally, the local listserv manager had tried some experiments,
sending "odd" files to the list but had not triggered any sequencing
discrepancies (at least, not on any of the messages that I requested).
In addition, he had notified the list of what he was trying and why,
so I consider it less likely than ever that someone intentionally
editted a message out of the archive.  I suppose the two-hour time
interval between the index and the request, though, isn't really the
relevant one, but rather the interval between the time the "missing"
message originally arrived at LISTSERV and the time of the request --
is that true?  I.e., any intervention during the time of accumulation
of the index-to-be would presumably cause a mis-sequencing.
 
Consider this scenario:  suppose there were a system crash (or maybe
just a shutdown) while a message was being processed.  The disk directory
is not updated immediately when a file has been modified, so it is
possible to update a file and still end up with the old version if CMS
is interrupted at just the right time.  Is it possible for LISTSERV to
lose a message from the archive in this way without losing the
corresponding entry in the currently accumulating index?
                                       John