Possible Solaris 8/INN-2.3.2 Problem
demarest at arraycomm.com
Fri May 25 16:38:24 UTC 2001
>>>>> "Alex" == Alex Kiernan <alexk at demon.net>
>>>>> wrote the following on 25 May 2001 09:57:48 +0100
Alex> James Ralston <qralston+ml.inn-workers at andrew.cmu.edu> writes:
>> On Fri, 18 May 2001, Timothy Demarest wrote:
>> > However, telnetting to port 119 or trying to run any newsreaders >
>> does not produce any results. The newsreaders timeout, the telnet >
>> attempt doesn't return the INN banner. Also, 'ctlinnd mode' hangs >
>> indefinitely, and the shuitdown process doesn't work. The innd >
>> process is running. There are no errors in the logs.
>> I'm quite familiar with running INN on Solaris systems, but I've never
>> encountered behavior like what you're describing. (I admit I haven't
>> used ovdb under Solaris, though.)
>> Try using truss(1) to follow innd, ctlinnd, et. al. as you run them.
>> If they're blocking on something, that should give you a good idea
>> where and what.
Alex> A lost wakeup in the pthread mutex stuff at a guess, I gave up and
Alex> switched to buffindexed, but that seems to have its own problems
Exactly! There seems to be problems with lightweight processes and the
mutex locking. I have inn totally disabled right now, so I wasn't able to
get the exact output per James' request. In fact, I can't even run
makehistory now to recereate the ovdb and history files.
I even tried recompiling everything under Solaris 2.5.1, but that hasn't
helped. If there doesn't seem to be a fix for 2.3.2, I'll probably go back
to 2.3.0 which works on my Solaris 8 box. In this case, what do I need to
do to recreated my ovdb and history files from the articles that are in my
spool? The history* and ovdb files currently are probably damaged pretty
badly - what ovdb files can I delete before recreting them?
Timothy Demarest ArrayComm, Inc.
demarest at arraycomm.com 2480 North 1st Street, Suite 200
http://www.arraycomm.com San Jose, CA 95131
More information about the inn-workers