yee@scam.Berkeley.EDU (Peter Yee) (09/06/90)
Wer recently converted news.arc.nasa.gov (USENET node ames) to run C news. The process was not without its trials and tribulations, but we managed to get it running. However, we are finding that C News is not processing batches fast enough to keep up with our incoming NNTP and UUCP neighbors. We run newsrun several times an hour (which isn't important, since the first one we invoke falls far behind the incoming batches and never exits). Relaynews seems to be chugging along, and articles come and go. Things just don't seem to happen fast enough. I've had to kill incoming NNTP service regularly since we aren't able to process all of our batches. And since the incoming NNTP articles are not processed fast enough, all of our NNTP neighbors end up transferring the article to us. Kind of like UUCP news! :-) B News always seemded to be able to handle the incoming load (then again, it could run multiple rnews invocations at the same time and it would accept incoming NNTP articles into the history file immediately, preventing our accepting the same article 20 times). At this point we are at the end of our rope. We can't afford (at least for this fiscal year) to purchase a faster machine. Our 90 megabyte incoming spool isn't getting much better (decreases slowly with NNTP off, increases at about 300k/minute with NNTP back on). Has anyone seen this news behaviour? Can anyone offer a solution? Please, we're desperate. -Peter Yee yee@ames.arc.nasa.gov ames!yee for Randy Robinson, Steve Schoch, and Peter Yee