Possible ovdb issues

[Brennon Church]

Hello all,

I've installed INN 2.3.0 on a system over here for testing.  Everything
seems to work fine until I allow people to start using it.  Once they
do, after a little time (anywhere from 10 minutes to 3 hours), the nnrpd
and innd process will not respond correctly.

To elaborate...  I'm running nnrpd separately (via the -D option), and
running innd on port 1919.  Whenever this error occurs, both ports 119 and
1919 (nnrpd and innd) will not respond when I try to connect to them.  They
accept the connection, but don't give the usual banner indicating that the
server is there.

Additionally, when I try to run nnrpd manually, it freezes.

I'm running this on a FreeBSD box.  I ran ktrace on nnrpd from the command
line, let it run for a bit, then took a look at the output.  It appears to
be a problem where it gets into a loop right after accessing the overview
data.  here is an excerpt from the tail end of the kdump output:

 10653 nnrpd    RET   close 0
 10653 nnrpd    CALL  open(0x80ee0c0,0x2,0x1b6)
 10653 nnrpd    NAMI  "/usr/local/news/spool/overview/__db.001"
 10653 nnrpd    RET   open 43/0x2b
 10653 nnrpd    CALL  fcntl(0x2b,0x2,0x1)
 10653 nnrpd    RET   fcntl 0
 10653 nnrpd    CALL  mmap(0,0x2000,0x3,0x201,0x2b,0,0,0)
 10653 nnrpd    RET   mmap 682979328/0x28b57000
 10653 nnrpd    CALL  close(0x2b)
 10653 nnrpd    RET   close 0
 10653 nnrpd    CALL  select(0,0,0,0,0xbfbff61c)
 10653 nnrpd    RET   select 0
 10653 nnrpd    CALL  select(0,0,0,0,0xbfbff61c)
 10653 nnrpd    RET   select 0
 10653 nnrpd    CALL  select(0,0,0,0,0xbfbff61c)
 10653 nnrpd    RET   select 0
 10653 nnrpd    CALL  select(0,0,0,0,0xbfbff61c)
 10653 nnrpd    RET   select 0

After this, the last two lines continue to repeat until I kill the nnrpd
process manually.

I've also run ktrace against the nnrpd processes running in the background,
as well as the innd process, only to see the same two lines repeating.

I'm not seeing anything in the logs that seem to be tied to this.  The
only unusual error I'm seeing is:

Jan 13 04:26:17 galanthis innd: free:-1 internal NCproc state 0

And it appears to be unrelated.

Has anyone else had issues like this?  I'm guessing that it has something
to do with the ovdb stuff, which I've just barely started playing with.  So
far the only way I can get things to start working again is to kill all
nnrpd, innd, and dbprocs processes and restart the whole thing.

Thanks in advance for any help you can offer.

-- 
Brennon Church
System Administrator
CTSnet Internet Services
mailto:ctsbac at cts.com
t 858.637.3600
f 858.637.361