🚀 go-pugleaf

tale@ten.uu.net (David C Lawrence) writes:

> If you are a
>C News site currently contemplating the INN switch on the basis of
>better NNTP performance, you might consider waiting just a little
>while longer and giving their NNTP a try.

'Tis ever thus. I installed INN about 10 hours before this posting.

With feeds set up as follows:

doc.ic.ac.uk\
	:*\
	:Tc,Wnm,Fnews.doc.ic.ac.uk\
	:/home/local/news/bin/nntplink news.doc.ic.ac.uk

I.e. the nntplink and inn "share" a spooling/batching file.

I'm seeing the following scenario, which occurs when there is some
delay sending articles to a remote site:

:r!pipe 'tail -400000c /news/logs/syslog | agrep doc'

Dec 17 10:50:36 pipe nntplink[18568]: news.doc.ic.ac.uk stats 100 offered 97 accepted 3 rejected 0 failed 1 connects
Dec 17 10:50:36 pipe nntplink[18568]: news.doc.ic.ac.uk xmit user 0.7 system 2.2 elapsed 777
Dec 17 10:51:15 pipe innd: frigate.doc.ic.ac.uk connected 17
Dec 17 10:51:29 pipe innd: frigate.doc.ic.ac.uk:17 closed seconds 14 accepted 4 refused 5 rejected 0
Dec 17 10:54:50 pipe innd: doc.ic.ac.uk:19:proc:18568 cant write Operation would block
Dec 17 10:54:50 pipe innd: doc.ic.ac.uk:19:proc:18568 blocked sleeping 120
Dec 17 10:57:52 pipe innd: doc.ic.ac.uk:19:proc:18568 wakeup
Dec 17 11:01:18 pipe innd: frigate.doc.ic.ac.uk connected 17
Dec 17 11:01:29 pipe innd: doc.ic.ac.uk:19:proc:18568 cant write Operation would block
Dec 17 11:01:29 pipe innd: doc.ic.ac.uk:19:proc:18568 blocked sleeping 120
Dec 17 11:03:08 pipe innd: frigate.doc.ic.ac.uk:17 closed seconds 110 accepted 1 refused 88 rejected 0
Dec 17 11:03:29 pipe innd: doc.ic.ac.uk:19:proc:18568 wakeup
Dec 17 11:03:29 pipe innd: doc.ic.ac.uk:19:proc:18568 cant write Operation would block
Dec 17 11:03:29 pipe innd: doc.ic.ac.uk:19:proc:18568 blocked sleeping 240
Dec 17 11:07:32 pipe innd: doc.ic.ac.uk:19:proc:18568 wakeup
Dec 17 11:07:32 pipe innd: doc.ic.ac.uk:19:proc:18568 cant write Operation would block
Dec 17 11:07:32 pipe innd: doc.ic.ac.uk:19:proc:18568 blocked sleeping 360
Dec 17 11:11:16 pipe innd: frigate.doc.ic.ac.uk connected 16
Dec 17 11:11:43 pipe innd: frigate.doc.ic.ac.uk:16 closed seconds 27 accepted 1 refused 67 rejected 0

Dec 17 11:13:57 pipe innd: doc.ic.ac.uk:19:proc:18568 cant flush count 5754 Operation would block
Dec 17 11:13:57 pipe innd: doc.ic.ac.uk dataloss 5754
Dec 17 11:13:57 pipe innd: doc.ic.ac.uk closed

Dec 17 11:15:23 pipe innd: doc.ic.ac.uk spawned doc.ic.ac.uk:19:proc:4068
Dec 17 11:21:17 pipe innd: frigate.doc.ic.ac.uk connected 32
Dec 17 11:21:31 pipe innd: frigate.doc.ic.ac.uk:32 closed seconds 14 accepted 1 refused 21 rejected 0
Dec 17 11:26:08 pipe nntplink[4068]: news.doc.ic.ac.uk stats 100 offered 70 accepted 30 rejected 0 failed 1 connects
Dec 17 11:26:08 pipe nntplink[4068]: news.doc.ic.ac.uk xmit user 0.4 system 1.1 elapsed 617

... and off we go again. It looks as if doc's news server just doesn't
want to listen for a while.

Am I missing something obvious here? Why isn't INN spooling?
Or is there something amiss with nntplink? (3.1.0)

Is it safe to make innd and nntplink share spooling/batching files like this?
If not, how should I join these two programs?

All help gratefully appreciated.

Ian
-- 
Ian Phillipps, Unipalm Ltd, 216 Science Park,		Phone +44 223 420002
Milton Road, Cambridge, CB4 4WA, England.		Phax  +44 223 426868
PIPEX is a division of Unipalm Ltd. - phone 0223 424616.