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

[McIDAS #ANL-325610]: imgdisp.k BAR arg sometimes give errors and WWDISP/FRNTDISP take a longer time to run.



Hi again Marty,

OK, the rebuild finished, and I reinstalled the newly built code:

<as 'mcidas'>
cd ~mcidas/mcidas2008/src
make clobber
make all
make install.all

I then stopped and restarted the LDM:

<as 'ldm'>
ldmadmin restart

Finally, I reran the test you sent along that was previously failing:

mcenv -f 13@600x800 -g 20 -i 48
imgdisp.k RTNEXRAD/N0R ID=MTX LINELE=230 230 PLACE=C MAG=1 1 EU=BREF SU=X ALL=1 
12 SF=YES BAND= X REFRESH='MAP FILE=OUTLUSAM MCOL=4 WID=1 GRA=(GRA);BAR (GRA) 
SU=X X COL=7 ORIENT=VER'
exit

This time there were no error messages from BAR (a good thing :-)

There are, however, some issues that I feel like I need to document:

- when running constructs like the 'imgdisp.k' invocation that includes
  a REFRESH= keyword sequence in a script, you must be aware that the
  commands run from the REFRESH= clause are run asynchronously --
  the MAP and BAR invocations will be run while the next imgdisp.k
  is running

  This is important since the script must keep running for as long as
  it takes for the MAP/BAR invocations to finish.  This may, in fact,
  require that the user include some sort of sleep in the McIDAS
  processing sequence (McIDAS command WAIT) so that the mcenv invocation
  doesn't exit before all of the other processing is done.

- there were three make-generated routines that failed during the re-build
  on your machine:

  kbprep.for
  nvprep.for
  mci_nvpp.for

  They all failed for the same reason, and that reason is still not
  100% clear to me yet.

  If you don't mind, I will use your machine as a testbed for
  investigating the problem (since it no longer occurs on our
  Solaris 10 x86 machine).

re:
> Also, I have noticed that the WWDISP/FRNTDISP commands are a lot
> slower than our other system. The WWDISP command takes about 30
> seconds each time it is run. I was wondering if these commands were
> pulling their information from the web somewhere or getting it local
> from our LDM.

After the rebuild/reinstall I started poking around in the 
/usr/local/ldm/data/mcidas
directory to see what I could see.  Here is what I found:

- the version of the wind profiler decoder that is being used is not
  deleting temporary files after decoding the broadcast data.  This
  left a LOT of .tmp_netcdf* files in /usr/local/ldm/data/mcidas.
  I deleted these files as 'ldm', and will look to upgrade the
  ldm-mcidas 'proftomd' decoder to a newer version.

- again, while looking through /usr/local/ldm/data/mcidas, I noticed
  that rapid access index files that should be updated by the XCD
  TEXT decoding were _not_ in the directory.  The net result of this
  is that access to data via the ADDE TEXT server (e.g., dataset
  RTWXTEXT) would be very slow since each request for data (like
  fronts and weather watch boxes) would result in a linear search
  through the daily repository files  for TEXT data, and those
  files are quite large:

-bash-3.00$ dmap.k \*.XCD
PERM      SIZE LAST CHANGED FILENAME     DIRECTORY
---- --------- ------------ ------------ ---------
-rw- 1261742320 Apr 23 17:59 DD091130.XCD /usr/local/ldm/data/mcidas
-rw- 1069854880 Apr 24 14:04 DD091140.XCD /usr/local/ldm/data/mcidas
2331597200 bytes in 2 files

  Apparently, the XCD setup BATCH file was never run as part of the
  XCD setup.  I did this for you as follows:

  <as 'mcidas'>
  cd $MCDATA
  batch.k XCD.BAT

  <as 'ldm'>
  ldmadmin restart

  Now the TEXT data that comes in will be indexed, and the index files
  will make access to the newly processed data much, much faster than
  without the index files.  Since the indexing does not go back in
  time, you will not see the benefit until after a few hours have
  passed.

Please let me know if/when you have questions.

Cheers,

Tom
--
****************************************************************************
Unidata User Support                                    UCAR Unidata Program
(303) 497-8642                                                 P.O. Box 3000
address@hidden                                   Boulder, CO 80307
----------------------------------------------------------------------------
Unidata HomePage                       http://www.unidata.ucar.edu
****************************************************************************


Ticket Details
===================
Ticket ID: ANL-325610
Department: Support McIDAS
Priority: Normal
Status: Closed