[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20010208: 20010206: Ldm quits unexpectedly (addendum)


Anne,

In reading through the below #2, as a suggestion in the future,
If a core file exists, you should be able to determine what program
created it with:
/local/ldm% file core
eg:
core:           ELF 32-bit LSB core file 80386 Version 1, from 'fips'


If the core file is coming from a program executed by pqact, then that should be
the name of the program in the output above. 

From John's answer below, it sounds like the core file was from 'rpc.ldmd'.

Steve Chiswell
Unidata User Support


>From: "John C. Nordlie" <address@hidden>
>Organization: UCAR/Unidata
>Keywords: 200102082255.f18MtBL23864

>Howdy.
>       Ok, I downloaded the ldm 5.1.2 binary for sunos_5.6-sparc,
>installed it, ran the scriptconfig (and ran it again after it complained
>about not having the qualified hostname), killed off all the rpc.ldmd
>and decoder processes, erased the queue, changed the runtime link,
>made a new queue, typed 'ldmadmin start' and ...
>
>it worked!
>
>No hangs, no core files, 'ldmadmin pqactcheck' returns proper syntax
>(it did before changing versions, btw), 'ldmadmin watch' shows data
>coming in (that worked before too), so all seems to be working right.
>Oh, our version of perl is 5.002.
>
>       So I can only guess there was some issue with the executable
>and the OS version, which is kind of weird but I'm not complaining!
>All I can say is a big 'Thankyou' for helping me cure the problem,
>and sorry for bugging you (but no promises that I won't in the 
>future :).
>
>Oh, and FYI, the core files were being generated from rpc.ldmd.
>
>I'll watch it like a hawk for a while, and let you know if anything
>strange happens.
>
>
>> Hi John, 
>> 
>> This is getting murky for me.  So, let me list the facts as I know them:
>> 
>> 1) Although 'ldmadmin start' won't complete, it will start rpc.ldmd and
>> pqact, which both continue to run.  But, rpc.ldmd and pqact run from the
>> command line appear fine, at least from the bit of output I've seen.
>> 
>> Something is keeping ldmadmin from completing this operation.  What
>> version of perl are you using?  
>> 
>> 2) Something is dumping core.  I can't help you much with this since I
>> don't know which program is doing it.  I suggest the pain-in-the-______
>> approach that I described before:  test individual lines in pqact.conf
>> one at a time.   Yeah, it's ugly - I feel your pain.  But, even though
>> it's slow and tedious it may be the fastest way in the long run.  For a
>> bit more information about this see
>> http://www.unidata.ucar.edu/cgi-bin/mfs/65/3581?79#mfs.
>> 
>> Btw, it's possible this could solve problem #1, above.
>> 
>> 3) Something is shutting down your decoders. Again, this could be solved
>> by #2, above.
>>  
>> But, the subject of your first email is "Ldm quits unexpectedly".  Is it
>> really the case that your ldm has stopped completely? 
>> 
>> 
>> > Rather than import the pqact.conf file and have the
>> > mailer munge it horribly, I'm putting a copy on my website
>> > at http://www.rwic.und.edu/~nordlie/pqact.conf so you can
>> > scold me about tabs and such.
>> > 
>> 
>> At first glance I see no problems.  But, looking for errors in
>> pqact.conf is one of my least favorable things to do.   Btw, have you
>> tried 'ldmadmin pqactcheck'?  That will find some problems, but not
>> all.  E.g., it may help you identify a problem, but a response of "no
>> problems" doesn't mean there aren't any.
>> 
>> It is very common for our users to take one of our big pqact.conf files,
>> edit it a bit, turn on the LDM for the first time with the result that
>> it's not working right.  Indeed, I am suspecting that most your problems
>> are due to an error in that file.  Again, although it's slow and
>> tedious, it may save time in the long run to start with a very small
>> pqact.conf file and make sure everything's working before adding another
>> entry.
>> 
>> 
>> > The os is solaris 2.5.1 (sunos 5.5.1), and it's the
>> > pub/binary/sunos_5.5-sparc directory on ftp.unidat.ucar.edu
>> > that doesn't have an ldm distro in it.  Yes, I have a compiler,
>> > I just didn't bother trying to build stuff after a quick
>> > attempt revealed I need to install a few other tools.  If that's
>> > the way I need to go, I can certainly do it, I'm just being lazy :).
>> > 
>> 
>> People here think that the 5.1.2 binary under built under SunOS 5.6
>> should work on your 5.5.1 system.  Since it's not too hard to install, I
>> suggest giving that a try.  At least, that may be easier than acquiring
>> the tools you need to build.
>> 
>> I would be interested in seeing your entire log file.  Can you make that
>> available to me like you did the pqact.conf file?
>> 
>> I'm sorry I can't tell you exactly what the problem is, but it's the
>> nature of a complex package that runs on many different platforms.  I
>> hope this is helpful.
>> 
>> Anne
>> -- 
>> ***************************************************
>> Anne Wilson                  UCAR Unidata Program            
>> address@hidden                      P.O. Box 3000
>>                                        Boulder, CO  80307
>> ----------------------------------------------------
>> Unidata WWW server       http://www.unidata.ucar.edu/
>> ****************************************************
>> 
>
>=========================================================================
>==)----------                   |                           ----------(==
>John Nordlie   N0RNB            |     Regional Weather Information Center 
>address@hidden            |              University of North Dakota 
>701-777-6112 / 701-777-3888 fax | PO Box 9007, Grand Forks, ND 58202-9007
>http://people.aero.und.edu/~nordlie/
>                                |
>     C Program.                 |       "If you ask me, most sports would
>     C Program Run.             |  benefit from an infusion of rocketry."
>     Run, Dammit, Run!          |                 -- Jason Fox, 'Foxtrot'
>==)----------      #include <std.disclaimer.h>              ----------(==
>=========================================================================
>