FC9 innd failing in libperl

[David L. Crow]

I upgraded from FC7 to FC9 and innd will not start.

When I run

   service innd start

It gives on "OK" indication, but innd does not start.  There is a 
message in /var/log/messages

   May 18 21:22:40 waterloo kernel: innd[27953]: segfault at 0 ip 
0020925b sp bfd15070 error 4 in libperl.so[12f000+26a000]

And the messages in /var/log/news/news.notice are

   May 18 21:23:39 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF001, len: 500.00  Mbytes, used: 495.00 Mbytes 
(99.0%)   1 cycles
   May 18 21:23:39 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF002, len: 500.00  Mbytes, used: 499.41 Mbytes 
(99.9%)   1 cycles
   May 18 21:23:39 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF003, len: 500.00  Mbytes, used: 6.42 Mbytes ( 
1.3%)   2 cycles
   May 18 21:23:39 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF004, len: 500.00  Mbytes, used: 4.16 Mbytes ( 
0.8%)   2 cycles
   May 18 21:23:40 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF005, len: 500.00  Mbytes, used: 497.14 Mbytes 
(99.4%)   1 cycles
   May 18 21:23:40 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF006, len: 500.00  Mbytes, used: 496.07 Mbytes 
(99.2%)   1 cycles
   May 18 21:23:40 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF007, len: 500.00  Mbytes, used: 491.14 Mbytes 
(98.2%)   1 cycles
   May 18 21:23:40 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF008, len: 500.00  Mbytes, used: 492.16 Mbytes 
(98.4%)   1 cycles
   May 18 21:23:41 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF009, len: 500.00  Mbytes, used: 494.72 Mbytes 
(98.9%)   1 cycles
   May 18 21:23:41 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF010, len: 500.00  Mbytes, used: 4.95 Mbytes ( 
1.0%)   2 cycles
   May 18 21:23:41 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF011, len: 500.00  Mbytes, used: 486.48 Mbytes 
(97.3%)   1 cycles
   May 18 21:23:41 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF012, len: 500.00  Mbytes, used: 494.13 Mbytes 
(98.8%)   1 cycles
   May 18 21:23:42 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF013, len: 500.00  Mbytes, used: 491.85 Mbytes 
(98.4%)   1 cycles
   May 18 21:23:42 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF014, len: 500.00  Mbytes, used: 6.99 Mbytes ( 
1.4%)   2 cycles
   May 18 21:23:42 waterloo cnfsstat[27999]: Class MCB for groups 
matching "*" Buffer BUFF015, len: 500.00  Mbytes, used: 0.20 Mbytes ( 
0.0%)   2 cycles
   May 18 21:25:53 waterloo innd: SERVER descriptors 1024
   May 18 21:25:53 waterloo innd: SERVER outgoing 1011
   May 18 21:25:53 waterloo innd: SERVER ccsetup control:14
   May 18 21:25:53 waterloo innd: SERVER lcsetup localconn:16
   May 18 21:25:53 waterloo innd: SERVER rcsetup remconn:3
   May 18 21:25:54 waterloo innd: controlchan! spawned 
controlchan!:19:proc:28217
   May 18 21:25:54 waterloo innd: innfeed! spawned innfeed!:20:proc:28219
   May 18 21:25:54 waterloo innfeed[28219]: ME starting innfeed 2.4.3 at 
Sun May 18 21:25:54 2008
   May 18 21:25:54 waterloo innfeed[28219]: loading /etc/news/innfeed.conf
   May 18 21:25:54 waterloo innd: serverfeed1.com opened 
earthlink.net:18:file
   May 18 21:25:54 waterloo innd: serverfeed2.com opened 
software.ibm.com:21:file
   May 18 21:25:54 waterloo controlchan[28217]: starting
   May 18 21:25:54 waterloo innd: serverfeed3.com opened 
microsoft.com:22:file
   May 18 21:25:54 waterloo innd: serverfeed4.com opened vmware.com:23:file
   May 18 21:25:54 waterloo innd: serverfeed5.org opened mozilla.org:24:file
   May 18 21:25:56 waterloo innfeed[28219]: ME source lost . Exiting
   May 18 21:25:56 waterloo innfeed[28219]: ME finishing at Sun May 18 
21:25:56 2008

(I replaced the actual feed names with generic names.)

I tried to debug innd using gdb, but got the following output

   Starting program: /usr/lib/news/bin/innd -d
   (no debugging symbols found)
   (no debugging symbols found)
   (no debugging symbols found)
   (no debugging symbols found)
   (no debugging symbols found)
   (no debugging symbols found)
   (no debugging symbols found)
   (no debugging symbols found)
   (no debugging symbols found)
   (no debugging symbols found)
   (no debugging symbols found)
   [Thread debugging using libthread_db enabled]
   (no debugging symbols found)

   Program exited with code 01.

and see the message

   May 18 21:20:28 waterloo innd: SERVER innd MUST be started with inndstart

in /var/log/news/news.crit, so I guess I won't be able to debug innd.

At this point, I am at a loss how to proceed with problem determination.

Any suggestions?
-- 
  David L. Crow                     Texas!  It's like a
  crow at OrangeBlood.org              whole other country.