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

20000127: McIDAS problems (cont.)

>From: Michael Voss <address@hidden>
>Organization: SJSU
>Keywords: 200001181849.LAA08980 McIDAS-X NLDNPLT

Mike,

re: try menu again and tell me what happens

I logged in this morning and see that a REDRECTion to UNIMENU.DEF that
I deleted yesterday was back.  Suspecting that somehow one rereads
REDIRECTions from LOCAL.NAM each time a session is started, I went
to the source and deleted the REDIRECTion for UNIMENU.DEF from LOCAL.NAM
(this was a mod in more recent versions of McIDAS).  Now, I can bring up
the MCGUI and display satellite images and overlay with surface data.
I notice, however, that you are not saving frontal analyses in a form
that would be usable by McIDAS.  I verified this by looking in
~ldm/etc/pqact.conf.  I went ahead and added the appropriate action in
pqact.conf and sent a HUP to pqact (as the user running the LDM).
Now frontal files usable by McIDAS should be put in the /data3/mcidas
directory.

>OK, I tried the menu and.....I was able to plot the current surface data. I
>was not however able to display current text data. The satellite and NLDN
>work fine. And the first problem I mentioned before is still there as you
>indicated above.
>
>BKGRUN {raobstn.gui SKEWT 94998 12 19 2000019  13 NO}                         
>BKGRUN: Done                                                                  
>        
>Error in startup script: syntax error in expression "12*(/12)"                
>        
>    while executing                                                           
>"expr 12*($time/12)"                                                          
>    (file "/export/home/mcidas/bin/raobstn.gui" line 81)     
>
>From the command line I can't get a skewT either:
>SKEWT 72493                                                                   
>SOUNDING UNAVAILABLE; NO SOUNDING RETURNED                                    
> -- SKEWT: Done

I finally traced this down to an old version of UNIDATA.MNU being
used.  Somehow, a previous version of UNIDATA.MNU is in your
~mcidas/data directory.  This version is not the one released with
McIDAS-X 7.60: it has not been ADDEized.  The Tcl/Tk support routines
distributed with 7.60 are, however, being used.  When you try to
display a sounding from the old menu, it is sending a time parameter to
the 7.60 raobstn.gui procedure in an unexpected form: HH instead of
HH:MM:SS.  The logic in the newer raobstn.gui tries to strip off the
':MM:SS' from a time specified in 'HH:MM:SS' format.  Since the Fkey
meny you are using sends only HH, the resultant string gets set to ' '
(blank.  This causes the Error in startup script: syntax error in
expression "12*(/12)" error you sent in above.

The question is now how/why you are attempting to use an old Fkey
menu.  As a sanity check for myself, I extracted the version of
UNIDATA.MNU contained in the ~mcidas/mcidasx760.tar.Z file on your
system.  I put this version of UNIDATA.MNU in the ~mcidas/tmp directory
and then took a look at it.  It is, in fact, the one expected for 7.60,
so you must have somehow copied an old version of UNIDATA.MNU over the
one in the new distribution.  Does this ring any bells?  Did you do
this because you had made a change in an old copy of the menu?

I verified that the 7.60 menu works as expected in plotting soundings
by doing the following from within a McIDAS-X session run as 'mcidas':

REDIRECT ADD UNIDATA.MNU "/export/home/mcidas/tmp
MCMENU UNIDATA.MNU

I then ran through displaying of images and plotting of several different
kinds of maps and all worked as expected.

>But maybe there really is no sounding... the files look a bit small:
>
>-rw-rw-r--   1 ldm      data       48536 Jan 27 15:16 /data3/xcd/MDXX0018
>-rw-rw-r--   1 ldm      data      112816 Jan 27 15:16 /data3/xcd/MDXX0028
>

Nope, it was due to the wrong Fkey menu being used.

>I will keep testing things here, but I thought I would just send that update.

Thanks.  There are some more things to attack.  I address some of thes
at the bottom of this email.

>I'm preparing for the first meeting of a lab class tomorrow morning, but we're
>not going to use the MCIDAS. They will be using it in Synoptic lab in the PM,
>but mostly the satellite imagery. So, don't worry if we can't get this
>resolved right away.

OK.

One of the next issues to be discussed is the access of GRID data files
for grid plots and cross sections.  The cross section routines in McIDAS
7.60 have not been ADDEized.  This means that the GRID cross section
routine XSECTG must still be run off of files that are accessible locally.
The problem with the interface in the Fkey and MCGUI menus is that they
were geared towards the GRID files that were in the Unidata-Wisconsin
datastream before July 1, 1999.  Now, one uses XCD to decode the NOAAPORT
model data into McIDAS compatible GRID files, but none of these looks
like the ones that came in the UW broadcast.

I did include, however, a utility routine called UWGRID in the 7.60
distribution that can be used to create GRID files from XCD created ones
that contain the same grids in the same order that they were in the
GRID files in the UW stream.  Since this conversion would need to
be kicked off by cron, I also included a shell script, uwgrid.sh,
that can be:

o edited (like xcd_run and batch.k) to set needed McIDAS environment
  variables (e.g., MCDATA, MCPATH, etc.)

o kicked off by appropriately timed cron entries to create the old
  GRID0001 and GRID0101 GRID files

If you are interested in doing this, you should read the infomation
at the beginning of uwgrid.sh (installed in the ~mcidas/workdata
directory) to see what you should do.

If you do not choose to recreate these GRID files, you can still use
the old Fkey menu to do horizontal plots of GRID fields. To do this
in the 7.60 menu, you have to change the string table definitions
of two strings: ?MODEL1 and ?MODEL2.  In the 7.60 version of UNIMENU.DEF,
these are defined as:

?MODEL1  "RTGRIDS/NGM-UW
?MODEL2  "RTGRIDS/MRF-UW

In order to use NGM GRIDs for the "North America - Horizontal" actions
in the 7.60 menu, you need to set ?MODEL 1 to RTGRIDS/NGM.  You
should probably do this by editing ~mcidas/data/UNIMENU.DEF.

To make the Fkey menu be able to display "Global - Horizontal" grids,
you should change the ?MODEL2 definition to be RTGRIDS/AVN.

I did both of the above in a running McIDSA-X session as 'mcidas' as follows:

TE ?MODEL1 "RTGRIDS/NGM
TE ?MODEL2 "RTGRIDS/AVN

and then verified the functionality of the 7.60 Fkey menu:

MCMENU UNIDATA.MNU
Model Output
F3     -> North America - Horizontal
F4     -> 700 mb Geopotential Loop

This worked fine.

In order to have all McIDAS-X users "get on the same page", you need to:

o install the real 7.60 version of the Fkey menu:

  <login as 'mcidas'>
  cd data
  mv UNIDATA.MNU UNIDATA.OLD
  mv ~/tmp/UNIDATA.MNU .

  After you do this, you should remove the REDIRECTion for UNIDATA.MNU
  from the user 'mcidas's environment:

  <start a McIDAS-X session>
  REDIRECT DEL UNIDATA.MNU

  Your session will still be able to find UNIDATA.MNU since you just moved
  the 7.6 version to ~mcidas/data and you have ~mcidas/data in MCPATH.

  Next, in order for users to use the Fkey menu to look at horizontal
  GRID plots, you need to: <edit UNIMENU.DEF and

  change:

  ?MODEL1  "RTGRIDS/NGM-UW
  ?MODEL2  "RTGRIDS/MRF-UW

  to:

  ?MODEL1  "RTGRIDS/NGM
  ?MODEL2  "RTGRIDS/AVN

o have each non-'mcidas' user do:

  <login>
  cd mcidas/data
  rm UNIMENU.DEF
  cd
  <start a McIDAS session>
  REDIRECT DEL UNIMENU.DEF       <- remove REDIRECTion for UNIMENU.DEF
  TENTER UNIMENU.DEF             <- will be found in ~mcidas/data by virtue
                                    of the user's MCPATH setting (if correct)

So, the last issue: why won't SFCRPT (and RAOBRPT, FOUSRPT, etc.) work.
There is something wrong with the TEXT server installation on vortex.
The reason I say this is because I can do the following:

DATALOC ADD RTWXTEXT ADDE.UNIDATA.UCAR.EDU
SFCRPT CA 1
K87Q 282052Z AUTO 21006KT 13/08 A3011 RMK AO2 SLP197
     T01280083 58017 PWINO TSNO=
KACV 282050Z 28008KT 20SM SCT200 13/05 A3014 RMK SLP207
     8/070 T00250054 50000=
KAPC 282054Z 09010KT 10SM BKN140 15/01 A3015 RMK AO2 SLP199
     T01500006 58015=
KBAB 282055Z 28003KT 50SM SCT200 13/05 A3017 RMK SLP216
     8/001 9/004 58017=
KBFL 282056Z 00000KT 6SM HZ CLR 14/06 A3012=
KBFL 282056Z 00000KT 6SM HZ CLR 14/06 A3012 RMK AO2 SLP199
 ...

I will continue to try and find out why your TEXT server isn't working
and get back to you when I figure it out.

Tom

>From address@hidden  Fri Jan 28 15:50:59 2000

Mike,

I just saw that a user named met171 is logged on multiple times and running
McIDAS-X.  This prompted me to take a look at ~met171/mcidas/data.  I
see that there is some cleanup that needs to be done there:

<login as met171>
<make sure no McIDAS session is running as met171>
cd mcidas/data
rm FRAMENH.001
rm SYSIMAGE.SAV
rm SYSKEY.TAB

The FRAMENH.001 file should only exist in a user's ~/.mctmp/nnnn directory.
It is a file that is designed to be removed at the end of a McIDAS session.
If it exists in any other directory in MCPATH, it will not be deleted and
can, therefore, cause problems if multiple sessions are being run from
the same account.

The file SYSIMAGE.SAV is no longer used in McIDAS-X.  It can be deleted
to save some space.

SYSKEY.TAB is needed for the Fkey menu and MCGUI. The one that is to be
used, however, will be the one that is being updated by the data decoders.
In a long previous version of McIDAS-X, SYSKEY.TAB needed to exist
in a user's McIDAS working directory, but that is no longer the case.

I see that /data3/xcd/STRTABLE keeps getting recreated (owned by ldm).
This should not be the case, so it is a hint at something running
incorrectly.  Yesterday afternoon, I moved it to the /data3/xcd/removed
directory, but I see that it is back:

ls -l /data3/xcd/STRTABLE
-rw-rw-rw-   1 ldm      data       12768 Jan 28 14:18 /data3/xcd/STRTABLE

We need to find out why this keeps getting recreated and stop it.

You are getting LOTS of email messages about your LDM setup.  Check out
the mail for the user running your LDM.

Tom

>From address@hidden  Fri Jan 28 16:17:15 2000

>I just got your message about met171 and ldm email, I'll check that out. I
>have some comments here from your previous work....

OK.

>running as 'mcidas':
>Well...I seem to be stuck in the same boat with the raobstn.gui? Oh
>wait, before I go further let me relate some info about UNIMENU. You
>asked below if I modified the menu. I did last summer, but only for
>local users, I did not change the UNI I made a LOCAL.MNU file which I
>had start up from .mcidasrc.

OK, this clears up what is going on!  The problem is that the the
old version of the menu, which you adopted for your LOCAL.MNU, doesn't
pass the time parameter in the form now expected by raobstn.gui.

>EG:
>
>     GOES-9 Western Infrared
>F  5 "EG 1 25;ALOOP #SYS(2113) #SYS(2114) 1 25 EC 32 130  -2 DWELL=2 7
>+ " MAP=YES H 9002 EU=IMAGE SU=IRTEMP LIMIT=32 BAR=#?BAR NO SF=YES
>
>I originally did this so I could visualize a 24 hour loop of just west
>coast satellite data from the menu.

OK, I understand now.

>And, If people wanted the regular
>menu they would enter MCMENU UNIDATA.MNU.

Got it.

>The was working fine before.
>Now I see that LOCAL.MNU file is in ~mcidas/workdata.

This would mean that only the user 'mcidas' can access it.  The
~mcidas/workdata directory should NEVER be included in any user's
MCPATH (except 'mcidas', that is).

>Hmm...I have
>~mcidas/workdata ahead of ~mcidas/data in my MCPATH and .....yep, check
>on the unidata web page and that looks like it's supposed to be that
>way...

Your MCPATH?  If you are not 'mcidas', then you need to remove
~mcidas/workdata from your MCPATH.  If you want to share any file(s)
among all users, like letting all users run your LOCAL.MNU menu, then
you should put it in the ~mcidas/data directory.

>Oh, I think I see....So, now when I start mcidas the F-KEY menu that
>comes up is the wrong/old one because it grabs LOCAL.MNU from
>~mcidas/workdata.

OK.

>But when I type MCMENU UNIDATA.MNU the new improved
>menu that you are refering to comes up. OK, I think I can handle that
>part of it. But, if I want to make a local menu that displays the 24
>hour loop of west coast imagery, do I need to edit the menu file as I
>did above, or is there a more elegant way to do that?

Actually, you can specify the number of images to put up in a loop through
the setting of string table values.  The only drawback is that it is
assumed that the number of images for a particular type of loop (like
an IR loop) will be the same regardless of the satellite it comes from.

Try the following:

<start a McIDAS-X session>
<bring up the UPC Fkey menu>

F7 - Menu Administration
F5 - Modify Menu System Parameters

change the setting for the Infrared Images Begfrm,Endfrm  ?IFRMS from
1 10 to 1 25.  Exit by "Update and Exit" so the new value is saved.
Then try putting up a loop.  The problem with this approach for the
new ADDE routines is that a loop will be setup of 25 frames regardless
of how many images you have.  Since you are keeping only 10 GOES-East
images, this will be bothersome for people looking at them.  What you
could do is change the string being used so that there is one for the
East and one for the West.  This would not be hard, but you would have
to do some "surgery" on the menu template.

re: REDIRECTion to UNIMENU.DEF removed
>Cool, thanks!

re: make sure about the 7.60 value of UNIDATA.MNU
>YES!, I addressed that above..

Super.

re: creating the GRID files that were once in the UW stream
>Awesome, I'll give it a try!

Let me know if you run into snags.

re: setting the dataset to access NGM and AVN GRIDs from the menu
>OK I will do this and keep this as a reference..

>thanks a million, I think we're getting there!

No problem.

Tom