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

20031124: Unidata McIDAS-X Version 2003 under Fedora at NIU (cont.)

>From: Unidata Support <address@hidden>
>Organization: UCAR/Unidata
>Keywords: 200311241913.hAOJDtEH002777 ldm-mcidas proftomd


On the off chance that the problem you are seeing is on weather2,
I SSHed from weather3 to take a look.  What I found was that none
of the ldm-mcidas decoding actions were running, and the ~ldm/logs/ldmd.log
file was full of pipe errors like you said.

I then found that the LDM was not running as the user 'ldm', but, rather,
as the user 'mcidas', and this was your entire problem.  So, I logged
on as 'mcidas' and stopped the LDM:

<login as 'mcidas'>
ldmadmin stop

and then logged on as 'ldm' and started the LDM:

<login as 'ldm'>
ldmadmin start

Now, the pnga2area decode processes are working correctly.

I see, however, that proftomd is failing:

Nov 24 21:02:18 proftomd[17038]: Starting up
Nov 24 21:02:18 proftomd[17038]: could not open /data/mcidas/.tmp_netcdf.QoJRvX
Nov 24 21:02:18 proftomd[17038]: Exiting

What is happening here is that the FSL wind profiler decoder is
trying to create a temporary file in /data/mcidas.  Since this directory
does not exist, the process was failing.  I changed the proftomd
entries in ~ldm/etc/pqact.conf:


# Hourly Summary
FSL2    ^FSL\.NetCDF\.NOAAnet\.windprofiler\.01hr
        PIPE    -close
        proftomd -vl /home/ldm/logs/ldm-mcidas.log -d /data/mcidas U2 WPRO 81
#####   PIPE    -close proftomd -d /home/data/mcidas U2 WPRO 81
# 6-minute
FSL2    ^FSL\.NetCDF\.NOAAnet\.windprofiler\.06min
        PIPE    -close
        proftomd -vl /home/ldm/logs/ldm-mcidas.log -d /data/mcidas U6 WPR6 91
#####   PIPE    -close proftomd -d /home/data/mcidas U6 WPR6 91

changed to:

# Hourly Summary
FSL2    ^FSL\.NetCDF\.NOAAnet\.windprofiler\.01hr
        PIPE    -close
        proftomd -vl /home/ldm/logs/ldm-mcidas.log -d /home/data/mcidas U2 WPRO 
#####   PIPE    -close proftomd -d /home/data/mcidas U2 WPRO 81
# 6-minute
FSL2    ^FSL\.NetCDF\.NOAAnet\.windprofiler\.06min
        PIPE    -close
        proftomd -vl /home/ldm/logs/ldm-mcidas.log -d /home/data/mcidas U6 WPR6 
#####   PIPE    -close proftomd -d /home/data/mcidas U6 WPR6 91

I then checked the integrity of the ~ldm/etc/pqact.conf file and sent
pqact a HUP signal telling it to reread the file:

ldmadmin pqactcheck
ldmadmin pqactHUP

Waiting around until profiler data came in shows that it is now being
decoded correctly:

Nov 24 21:08:44 proftomd[17952]: Starting up
Nov 24 21:08:44 proftomd[17952]: Decoding 2003328.2054 data into 
Nov 24 21:08:44 proftomd[17952]: WARNING: MD #cols 35 < dataset #stn 35
Nov 24 21:08:44 proftomd[17952]: Exiting

So, the file is being written now.  The warning being issued by proftomd
is a separate issue that I will need to look into further.

I then noticed that a lot of the data files in /home/data/mcidas had
not been scoured in quite awhile.  I traced this down to the following:

1) the McIDAS scouring was being done (through a cron job running mcscour.sh)
   by 'mcidas'  --  this is not bad, but
2) the output log file for the scouring, /home/mcidas/workdata/scour.log,
   was owned by 'ldm' and was not writable by 'mcidas'

I commented the cron initiation of mcscour.sh in the 'mcidas' account
and added the same entry to the 'ldm' account.  In doing this, I

1) made a copy of /home/mcidas/workdata/mcscour.sh in /home/ldm/decoders
2) added a cron entry to run that copy of mcscour.sh
3) added a cron entry to rotate ldm-mcidas log files

I think things are running correctly now.  Please let me know if you see
strange things.


>From address@hidden Mon Nov 24 15:30:17 2003

re: LDM was running as 'mcidas'

>Oh, boy. Sometime, I must have, in a bunch of confusion setting everything 
>up, restarted LDM as user McIDAS. Cute. Sorry about that.

re: the profiler data file is now being written

>Great! Whew. OK, next problem...
re: setup scouring in 'ldm' account and added rotation of ldm-mcidas log files

>Thanks again for all your help! Actually, they are being strange.

>I bring up the plot/contour menu, it crashes before I can do anything. It 
>just hangs the moment I bring the menu up.

>From address@hidden Mon Nov 24 15:31:10 2003

>SAO/Metar Plots/Contours menu, sorry. I just did it now and it is hung 
>hard on weather2.

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

NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.