20021007: Proftomd hanging on RedHat 8.0 (cont.)

>From: Gilbert Sebenste <address@hidden>
>Organization: NIU
>Keywords: 200210050242.g952g0127088 ldm-mcidas proftomd

Gilbert,

I have been logged onto weather2 for the past couple of days trying
to troubleshoot the problems you have been having running the ldm-mcidas
FSL wind profiler decoder, proftomd.  Nothing that I see makes any sense!
I can't find any obvious bugs in proftomd, so I can't say for sure
why it hangs and sucks up CPU.

I did, however, figure out a bandaid kludge to get proftomd working
again.  It is not a fix; it is a kludge!  So, proftomd is once again
decoding profiler data into McIDAS MD format.

The second problem you saw and reported to the ldm-users email list was
with the McIDAS-XCD synoptic decoder, DMSYN (dmsyn.k).  I have been
trying to find out why/where it apparently goes into an infinite loop,
but I have not yet been successful.  I rebuilt dmsyn.k using the debug
flag and havn't seen it mess up since (scary).  I will continue to let
it run hoping that it will go into an infinite loop while I am watching.
If/when it does, it needs to be killed by sending it an ABRT signal:

kill -ABRT <pid_of_dmsyn.k>

Hopefully, this will cause it to dump out a core file that I can then
examine.  In order to allow this, I set the coredumpsize in the
'ldm' environment to unlimited:

<logon as 'ldm'>
unlimit coredumpsize

and then stopped and restarted the LDM:

ldmadmin stop
<wait for all LDM processes to exit>
ldmadmin start

If the LDM is stopped and restarted by 'ldm' in a different session, the
core file size will return to nothing and failures will not dump out
core files.

Tom

>From address@hidden Thu Oct 17 21:06:08 2002
>Subject: Re: 20021007: Proftomd hanging on RedHat 8.0 (cont.)

Hi Tom,

re: bandaid kludge to get proftomd working

>Thanks. FYI, I did install the latest kernel on weather2 this afternoon. 
>See what happens now. I haven't gotten to weather yet.
 
re: investigating synoptic decoder

>OK, will do. See if the kernel patch helps. Thanks for doing this! Dan 
>Vietor was the first to point out the kernel's shortcomings; hopefully, 
>this is fixed now.

*******************************************************************************
Gilbert Sebenste                                                     ********
Internet: address@hidden    (My opinions only!)                     ******
Staff Meteorologist, Northern Illinois University                      ****
E-mail: address@hidden                                 ***
web: http://weather.admin.niu.edu                                      **
Work phone: 815-753-5492                                                *
*******************************************************************************