Re: pqing stopped inserting when downstream requested feed

        I would love to take credit for altruism. But, in truth I can't.
I am upgrading machines here and stumbled on the same thing.
At 09:19 PM 2/14/2001 +0000, you wrote:
>The FreeBSD box is AMD.  The Solaris 7 box -- not 8 (my mistake) is Intel.
>James, thanks for all your work reproducing our problem.
>Jessica M. Thomale
>Oklahoma Climatological Survey
>E-mail: jthomale@xxxxxx
>Mail: 100 E. Boyd, Suite 1210 Norman, OK 73019-1012
>Phone: (405) 325-7809
>Fax: (405) 325-2550
>> From: "James D. Marco" <jdm27@xxxxxxxxxxx>
>> Date: Wed, 14 Feb 2001 16:06:31 -0500
>> To: Jessica Thomale <jthomale@xxxxxxxxxxxxxxxxxx>
>> Cc: ldm-users@xxxxxxxxxxxxxxxx, support@xxxxxxxxxxxxxxxx
>> Subject: Re: pqing stopped inserting when downstream requested feed
>> Jessica,
>> I use the ldmd.conf and "REQUEST" from a local server.
>> This is all IDD, but I can now confirm your problem.
>> a) I updated one of my systems to Red Hat Linux 7.0
>> b) Updated the LDM and recompiled
>> c) Configured for a basic WMO ingestion, no decoders
>> d) Removed the old queue and made another (256meg)
>> e) Started ldm
>> f) It killed the data stream on the upstream machine 
>> This was working before the upgrade to RH7.0.  The server
>> and client were communicating OK. The data stream was OK, at least
>> for the small WMO stuff. Both machines are Intel.
>> The server is still using ldm5.1.2 on RH6.2.
>> The client is RH7.0 (now)
>> You appear to have trouble with:
>> FreeBSD 4.2 server
>> Solaris 8 
>> I am assuming these are both Intel(?).
>> The errors I get are similar to yours. I am attaching the full logs,
>> but be aware that the client log had MY mistake in it.  I  had
>> forgotten to build a new queue before starting the ldm (dumb).
>> Anne, 
>> There is a pretty clear relationship between starting the
>> client, and crashing a server process. I did this a couple of times
>> to be sure. Any how, some surface thoughts follow. You are probably
>> already looking, but:
>> Have any of the file locking semantics changed?
>> Or the socket interface? (AF_INET, AF_UNIX, AF_LOCAL)
>> Have any new functions been added or changed in the queuing?
>> (search, indexing, qinsert(), qdelete())
>> Anyway, I uninstalled ldm-5.1.3, reinstalled ldm-5.1.2 and it is
>> working again.  The server configuration was unchanged, except to
>> stop and restart.  I have ldm-5.1.3 running on a RH6.2 box. It is
>> not crashing things.
>> jdm
>> At 02:01 AM 2/14/01 +0000, you wrote:
>>> Good Evening,
>>> We are requesting a NOAAPort feed using pqing -P <port number> on a ldm
>>> server that serves this data to downstream ldm hosts.
>>> Today, when the downstream ldm server started requesting a feed from the
>>> upstream ldm server, one of the pqing processes (importing the NWSTG
>>> channel) mysteriously died. I attached a copy of the two ldmd log files.
>>> Both machines are running ldm-5.1.3. The upstream server is a FreeBSD
>>> (version 4.2) box.  The downstream server is a Solaris/x86 box (version
>>> There appears to be a problem with the product queue on the upstream ldm
>>> server.
>>> Has anyone experienced any problems similar to this problem?
>>> Thank you in advance for any assistance,
>>> Jessica
>>> -- 
>>> Jessica M. Thomale
>>> Oklahoma Climatological Survey
>>> E-mail: jthomale@xxxxxx
>>> Mail: 100 E. Boyd, Suite 1210 Norman, OK 73019-1012
>>> Phone: (405) 325-7809
>>> Fax: (405) 325-2550
>>> Attachment Converted: "c:\program files\bear
>> access\winba\eudora\attach\ldmd.log.downstream"
>>> Attachment Converted: "c:\program files\bear
>> access\winba\eudora\attach\ldmd.log.upstream"
James D. Marco, jdm27@xxxxxxxxxxx, jmarco1@xxxxxxxxxxxx
Programmer/Analyst, System/Network Administration, 
Computer Support, Et Al. 
Office:                 1020 Bradfield Hall, Cornell University 
Home:                   302 Mary Lane, Varna    (607)273-9132
Computer Lab:   1125 Bradfield          (607)255-5589