On Mon, 6 Jan 2003, Support AK wrote:
> Hi, can you send me the status or email you get when it jams, and your
> dnews.conf, thanks.
Here is my dnews.conf... comments removed.
------------------------------------------
myname www.dandynews.com
nntp_feeder paxfeed.eni.net
diskspace 535555
spool /big/spool
workarea /big/work
history /big/history
xover_dir /big/xover
manager [EMAIL PROTECTED]
timezone NZST
mail smtp
df df -k
sendmail /usr/lib/sendmail -t
confirm *dandy.net
too_old 35
loglvl info
debug_rotate 10000000
life 20
update_at 0,10,20,30,40,50 *
expire_at 59 3
expire_any true
config /usr/local/dnews
logdir /var/log/dnews
workarea_min 10
history_min 50
spool_min 50
post_rec true
suck_requeue true
access_forward true
dns_check true
slave_n 9
max_speed 100000000000
cache_onxover true
post_rec true
autoexpire 95
cachedall true
concurrent_limit 200
head_cache 10000
max_groups 100000
max_out 20000
fts_dir /big/spool/fts
slave_restart 5
spam_allowgroups local.*,dandy.*
log_rotate_nostart true
bucket_size 1000000000
auth_case true
use_smtp true
------------------------------------------
> And platform type and dnews version. Also physical memory you have installed
> and swap size.
Platform is Linux... 2.4.19-grsec... GLIBC 2.2.5 patched for libbind bug.
It will be 2.4.20-grsec soon. I've had the problem with various 2.4
kernels going back to 2.4.6, IIRC.
Dnews is 5.6e3... again, I can recell having the problem and upgradeing
from the 5.5g days.
I have 1 Gig of real memory, 1 Gig of swap...
total: used: free: shared: buffers: cached:
Mem: 925278208 920817664 4460544 0 36888576 452673536
Swap: 1052794880 16158720 1036636160
MemTotal: 903592 kB
MemFree: 4356 kB
MemShared: 0 kB
Buffers: 36024 kB
Cached: 431696 kB
SwapCached: 10368 kB
Active: 153068 kB
Inactive: 708540 kB
HighTotal: 0 kB
HighFree: 0 kB
LowTotal: 903592 kB
LowFree: 4356 kB
SwapTotal: 1028120 kB
SwapFree: 1012340 kB
> Also next time it occurs, copy expire.log and send that too. (but send the
> other stuff first, I may be able to find a clue without that)
It just happened last night... unfortunately, the expire file had grown to
2.1 Gigs and was truncated. I /dev/nulled it, but by the time this
happens again, I'm sure it will be huge.
I have the status email. I'll forward that in the next email...
> At 03:40 AM 1/5/03 -0500, you wrote:
> >Hello.
> >
> >I've been having this problem for long time now, over several Dnews
> >versions.
> >
> >Sometimes, about twice a month, the expire seems to fail, and a pile goes
> >considerably over it's limit. Then the server stops taking news.
> >The logs show nothing unusual. It can happen at any time of the day.
> >
> >I have a 500 Gig spool, and an expire.conf like this:
> >
> >pile 1 use 25% groups *
> >pile 2 use 65% groups *binaries*,*warez*,*pictures*,*sounds*,*mp3*
> >pile 3 days 365 groups local.*,dandy.*
> >pile 4 days 1 use 1% groups control*
> >
> >A happy server is usually 132G in pile 1, 346G in pile 2.
> >Pile 3 and 4 are neglible. I have about 14 Gig for overflow.
> >Unfortunately, pile 2 will sometimes hit 67 or 68%, and the whole works
> >stops. My buckets are 1 Gig max. I had 1800 of them last I checked.
> >__
> >I had previously reduced the piles, but even with pile 2 at
> >45% and a 120 gig overflow, it still did.
> >
> >This problem is driving me nuts and I got to get it solved.
> >This server gets fed by Cidera and sustains about 23 MBps,
> >so it takes very little time to malfunction. I guess the 346 Gigs
> >of binaries has a retention of 2-3 days... and this is after Cidera
> >started filtering out alot of crap, and I started spam filtering.
> >
> >Thanx in advance for any help you can provide.
>
>
----Steve
Stephen Amadei
Dandy.NET! CTO
Atlantic City, NJ
