🚀 go-pugleaf

I am running nn version 6.3.10 on a Convergent Megaframe (sVr2).

When nnmaster is collecting articles, it sometimes writes a 0 length
data file (DB_DIRECTORY/DATA/nnn.d) for a group that has active
articles.  Needless to say, nn won't read any articles in these
groups.

For these 'broken' groups, the M)aster G)roup command in nnadmin shows
something like this:

news.software.nn	348
first/last art: 000016 000044
   active info: 000016 000053
Offsets: index->0, data->0

The problem is not consistent; if I run nnmaster -I several times, it
will trash different groups each time.

I have tried to step through nnmaster with sdb, but I haven't been
able to catch it screwing up yet.

Any ideas???

BTW... nn 6.3.6 ran just fine.

-- Bob Finch
   bob%csispt.UUCP@unicorn.wwu.edu
   +1 206 671-6214