[uw.mfcf.software] dcsu news

wcwince@watmath.waterloo.edu (09/01/89)

From: "Bill Ince [MFCF]" <wcwince>

deleted a very large number of 'rnews -U' and 'sendbatch' entries
in the news queue.
rm'd most of the files in /usr/spool/news/.rnews,
as they seemed to be duplicate news batches
(there were hundreds).
Not clear what was/is causing the buildup.

rjwhite@watmath.waterloo.edu (RJ White) (09/01/89)

In article <28809@watmath.waterloo.edu> wcwince@watmath.waterloo.edu writes:
	From: "Bill Ince [MFCF]" <wcwince>
	
	deleted a very large number of 'rnews -U' and 'sendbatch' entries
	in the news queue.  rm'd most of the files in /usr/spool/news/.rnews,
	as they seemed to be duplicate news batches (there were hundreds).
	Not clear what was/is causing the buildup.

The billions of batched 'rnews -U' and 'sendbatch' commands were from
the cron every 15 minutes or so that couldnt run because sendbatch on
maytag was busy filling /usr/spool on many machines with billions of
identical batches.
This was because the cron runs sendbatch and when sites backfeed news, they
do a 'rsh maytag .... sendbatch' and the news locking stuff in sendbatch
doesnt work on maytag because the bourne shell is broken there.
sendbatch changed and installed on maytag for testing to get around this.

The bug is that error statuses are not returned properly when doing a
	if pid=`flock  ...   `
illustated by :

$ grep -s tty /etc/ttys
$ echo $?
0
$ grep -s nosuchstr /etc/ttys
$ echo $?
1
$ `grep -s tty /etc/ttys`
$ echo $?
0
$ `grep -s nosuchstr /etc/ttys`
$ echo $?
0			<- this should be 1

if theres a place to send bug reports to Sequent, feel free to send it in...

Now if only exclusive flock's would work on MIPS....