Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

INND Woes...

0 views
Skip to first unread message

Wilbur E. Coyote

unread,
Mar 13, 1998, 3:00:00 AM3/13/98
to

This is a multi-part message in MIME format.
--------------812DAA641D848117FD7D32BB
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Hello all!

Help with either of the 2 listed problems is greatly appreciated.

=======================================================================
Configuration:

Running INND 1.7.1 with patches on Digital UNIX 4.0

OSF1 web.onecall.net V4.0 564 alpha


I'm running with approximately 8GB RAID 5 on a Winchester driver
array

df -k
Filesystem 1024-blocks Used Available Capacity
Mounted on
root_domain#root 131072 74722 55200 58% /
/proc 0 0 0 100%
/proc
usr_domain#usr 1048576 771407 277169 74%
/usr
/dev/rz2c 1985982 49513 1737870 3%
/sys_xpnd
/dev/rzb2c 992735 371087 522374 42%
/news_ctl
/dev/rzc2c 7698488 5999050 929589 87%
/news_arts

My primary purpose is NNTP relay, I server a limited number of
NNRP requests.


==============================================================================
Problems
----------------------
1. I'm running a High-water-mark/Low-water-mark space based
expire program with simple warp acceleration to meet diskspace
requirements. Sometimes during my expire process, my news.daily
cron job kicks in. I wind up with an "Out of memory!" error in
my "/var/log/news/errlog" file. I'm running 384MB RAM and 855
swap.

kern.debug:Mar 6 15:47:43 web vmunix: physical memory = 384.00
megabytes.
kern.debug:Mar 6 15:47:43 web vmunix: available memory = 365.17
megabytes.
kern.debug:Mar 6 15:47:43 web vmunix: using 1467 buffers
containing 11.46 megab
ytes of memory

vmstat -s
Virtual Memory Statistics: (pagesize = 8192)
3256 active pages
3801 inactive pages
10613 free pages
6487 wired pages
15972355 virtual memory page faults
3622116 copy-on-write page faults
4216435 zero fill page faults
393407 reattaches from reclaim list
3793662 pages paged in
2460 pages paged out
359315319 task and thread context switches
227660776 device interrupts
385407383 system calls

sbin/swapon -s
Swap partition /dev/rz0b (default swap):
Allocated space: 49125 pages (383MB)
In-use space: 5838 pages ( 11%)
Free space: 43287 pages ( 88%)

Swap partition /dev/rz0h:
Allocated space: 60324 pages (471MB)
In-use space: 4328 pages ( 7%)
Free space: 55996 pages ( 92%)


Total swap allocation:
Allocated space: 109449 pages (855MB)
Reserved space: 24130 pages ( 22%)
In-use space: 10166 pages ( 9%)
Available space: 85319 pages ( 77%)
bash#

I suspect a memory leak in expire or in INND itself but can find
no evidence. Anyone with previous experience or suggestions on
tracking down this problem is welcome.


--------------------
2. I'm running into a nearly 4 hour expire time which seems
excessive. Does anyone know a way to improve my expire times?
My present expire command looks like:

/var/news/bin/expire -p -v1 -l -n -w 0

============================================================================

Any help is greatly appreciated.

Please reply to wco...@onecall.net or tnor...@onecall.net

Regards,

Troy W. Norcross
USENET admin at One Call Communications

Cheers!

.t.
--------------812DAA641D848117FD7D32BB
Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Wilbur E. Coyote
Content-Disposition: attachment; filename="vcard.vcf"

begin: vcard
fn: Wilbur E. Coyote
n: Coyote;Wilbur E.
org: ACME, INC.
email;internet: wco...@onecall.net
title: Head Flunky
note: Friends of: Bill W. and Dorothy.
x-mozilla-cpt: ;0
x-mozilla-html: FALSE
version: 2.1
end: vcard


--------------812DAA641D848117FD7D32BB--


0 new messages