Re: 20010426: HELP Gilbert

Unidata Support wrote:
> 
> ------- Forwarded Message
> 
> >To: General Support <address@hidden>
> >From: Gilbert Sebenste <address@hidden>
> >Subject: HELP! I need it bad!!!
> >Organization: UCAR/Unidata
> >Keywords: 200104262115.f3QLFCL26765
> 
> My machine, weather.admin.niu.edu, just crashed hard for the second day in
> a row, right around 19Z (3 PM CT). I looked in /var/log/messages, ldmd.log
> file.../var/log/secure...nada. Not a clue. Everything was alright right up
> to the point when it crashed. And when it does, it does very quickly,
> requiring an e2fsck that has to go through twice to fix all of the inodes
> on my data directory, /dev/sda1, or /home/data. IE, I suspect this is an
> LDM problem.
> 

Hi Gilbert,

Dang - that sounds unpleasant!

> weather3.admin.niu.edu hasn't crashed yet, but it is not saving quite as
> much as weather is. The only thing I can figure out are two things:
> 
> 1. something is wrong with the LDM (I haven't recompiled it, I compiled it
> under Redhat Linux 6.2 and am still using it now under 7.1), or
> 

I would definately try recompilation due to the change in the kernel. 
Did you make a new queue?  If not, that would be worth trying.  We do
have the LDM running on sites with version 7.0.  


> 2. My cheap NEXRAD clean script, which just basically does a rm -r at 3:45
> AM every day to my 88d directory, is causing problems that show up almost
> exactly 12 hours later.
> 
> No security breaches are evident. Yet, although I am not sure, it might
> have crashed at almost the same time or the same time as it did yesterday.
> I only have a WXP metar conversion at 3 PM, if I recall here, that I do
> out of cron at 3 PM. And, this didn't happen until I got RH 7.1 installed.
> 
> Any ideas greatly welcomed...yes, all 7.1 patches are on. I will be
> unavailable until 10 PM tonight, but if you have any ideas, suggestions,
> criticisms, etc, drop me an email. And oh yeah, feel free to log in. Anne,
> I believe, has my ldm password. So does Tom Y.
> 
>

I logged in and looked around on weather.  The only mildly suspicious
thing I saw in the log was a connect just before the last crash from
weather3 for UNIDATA and MCIDAS feeds at Apr 26 19:58:29.  It might be
interesting to see what weather3's log says at that time.  But, I see an
identical connection occuring on Apr 27 that has apparently caused no
problems...

I would try recompiling the ldm and building a new queue, an easy test
to make.  Let me know what happens.

Anne

*******************************************************************************
> 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                                                *
> *******************************************************************************
> 
> ------- End of Forwarded Message

-- 
***************************************************
Anne Wilson                     UCAR Unidata Program            
address@hidden                  P.O. Box 3000
                                  Boulder, CO  80307
----------------------------------------------------
Unidata WWW server       http://www.unidata.ucar.edu/
****************************************************