20011227: SSEC McIDAS not able to get data from Unidata McIDAS 7.804

>From: Robert Mullenax <address@hidden>
>Organization: NMSU/NSBF/Universal Weather
>>Keywords: 200112271446.fBREkPN17596 McIDAS-X ADDE gcc/g77

Robert,

>We have been having problems with the socket based data feed between
>our NOAAPORT SDI and our SunFire 4800 where our XCD decoders reside
>(SSEC McIDAS 7.8).  They will just stop talking to each other.  I had
>wanted for some time to switch to the LDM since we are running an LDM
>anyway to feed GEMPAK.  The easiest way for me to test that was to do a
>Unidata McIDAS install since it is LDM ready right out of the box.  So
>I grabbed 7.804 and put it on a Solaris Intel box we have here and
>built it with gcc/g77 with your instructions.  The build went off
>without a hitch and I got everything running (XCD..etc) in short
>order.

Sounds good so far...

>I used my workstation which is running Unidata McIDAS 7.7 to
>point at it and get data..everything worked great.  However, when I
>tried to point at it from the boxes that have SSEC McIDAS (either 7.705
>or 7.8) they do not see any of it's datasets.

Wow, this is totally unexpected!

Is there a chance that you logged onto the SSEC McIDAS sessions and
then tried to access the Unidata ADDE server built using gcc/g77?

Conversly, what happens when you try to access an ADDE server on the
machine running SSEC McIDAS from a box running the Unidata code
built with gcc/g77?

>For example, if I do a
>DATALOC ADD RTWXTEXT IMPACTWX12.UNIV-WEA.COM (that's the x86 box) and
>then do DSINFO X RTWXTEXT I get:
>
>$ dataloc.k ADD RTWXTEXT IMPACTWX12.UNIV-WEA.COM
>
>Group Name                    Server IP Address
>--------------------         ----------------------------------------
>RTWXTEXT                     IMPACTWX12.UNIV-WEA.COM
>
><LOCAL-DATA> indicates that data will be accessed from the local data
>directory.
>DATALOC -- done
>$ dsinfo.k X RTWXTEXT
>    No Datasets found of Type: IMAGE in Group: RTWXTEXT
>    No Datasets found of Type: POINT in Group: RTWXTEXT
>    No Datasets found of Type: VPROF in Group: RTWXTEXT
>    No Datasets found of Type: GRID  in Group: RTWXTEXT
>    No Datasets found of Type: TEXT  in Group: RTWXTEXT
>    No Datasets found of Type: NAV   in Group: RTWXTEXT
>DSINFO -- done
>$ 

Hmm...  Unfortunately, DSINFO does not support the TRACE= keyword, so
we can't use it to have the server dump debug information.
IMGDISP/IMGLIST, on the other hand do support TRACE=.  Given this, what
kind of debug output is found after doing an IMGLIST to a machine
running Unidata McIDAS-X 7.804 built with gcc/g77 from a box running
SSEC McIDAS?

I think we can learn alot by comparing the 'trce' file created by the
agetserv server on the box running Unidata McIDAS 7.804 when contacted
by the different clients.  Without this debug output, I am at a loss
for what could be going on...

>However, if I point at one of the NSBF machines which is running
>Unidata 7.803 it works fine:

So, from a machine running an SSEC McIDAS-X session, you are able to
get ADDE information back from Unidata McIDAS-X 7.803 built without
using g77?  If this is true, it is very interesting and most likely
indicates that there is a lingering problem in one or more McIDAS
routines.

>$ dataloc.k ADD RTWXTEXT ULDBLDM.NSBF.NASA.GOV
>
>Group Name                    Server IP Address
>--------------------         ----------------------------------------
>RTWXTEXT                     ULDBLDM.NSBF.NASA.GOV
>
><LOCAL-DATA> indicates that data will be accessed from the local data
>directory.
>DATALOC -- done
>$ dsinfo.k X RTWXTEXT
>    No Datasets found of Type: IMAGE in Group: RTWXTEXT
>    No Datasets found of Type: POINT in Group: RTWXTEXT
>    No Datasets found of Type: VPROF in Group: RTWXTEXT
>    No Datasets found of Type: GRID  in Group: RTWXTEXT
>
>        Dataset Names of Type: TEXT  in Group: RTWXTEXT
>
>Name         NumPos   Content
>------------ ------   --------------------------------------
>FOUS14           1    Real-Time FOUS14 Text
>SFCHOURLY        1    Real-Time SFC Hourly Text
>SYNOPTIC         1    Real-Time SYNOPTIC Text
>TERMFCST         1    Real-Time Terminal Forecast
>UPPERAIR         1    Real-Time Upper Air Text
>
>    No Datasets found of Type: NAV   in Group: RTWXTEXT
>DSINFO -- done

OK.

>If I do the same from my Universal workstation running Unidata's 7.704 it
>works fine:
>
>/WX/users/rrm% dataloc.k | grep IMPACTWX12
>RTNEXRAD                     IMPACTWX12.UNIV-WEA.COM
>RTPTSRC                      IMPACTWX12.UNIV-WEA.COM
>RTWXTEXT                     IMPACTWX12.UNIV-WEA.COM
>
>/WX/users/rrm% dsinfo.k X RTWXTEXT
>    No Datasets found of Type: IMAGE in Group: RTWXTEXT
>    No Datasets found of Type: POINT in Group: RTWXTEXT
>    No Datasets found of Type: VPROF in Group: RTWXTEXT
>    No Datasets found of Type: GRID  in Group: RTWXTEXT
>
>        Dataset Names of Type: TEXT  in Group: RTWXTEXT
>
>Name         NumPos   Content
>------------ ------   --------------------------------------
>FOUS14           1    Real-Time FOUS14 Text
>SFCHOURLY        1    Real-Time SFC Hourly Text
>SYNOPTIC         1    Real-Time SYNOPTIC Text
>TERMFCST         1    Real-Time Terminal Forecast
>UPPERAIR         1    Real-Time Upper Air Text
>
>    No Datasets found of Type: NAV   in Group: RTWXTEXT
>DSINFO -- done
>/WX/users/rrm% 

Weird!

>I even installed a Unidata McIDAS version on the same box where the
>SSEC version is installed and it would not work.  Of course the
>mcinet's might be slightly different and the user mcadde is set up
>differently.

Let's not go there just yet.

>I just thought this might be of interest to you as it involves your
>latest release and the new gcc/g77.

Yes, absolutely!

>Are there any more ADDE diagnostic
>commands that will let me see more of what is going on?

We need to do an IMGLIST while pointing to the machine running Unidata
McIDAS-X 7.804 and make sure to include the TRACE=1 keyword sequence on
the command line.  You should end up with a file named 'trce' in the
~mcidas/mcidas/data directory on the server.  If you do NOT end up with
this file, we might have found the problem (one of setup).  If you do
find this file, you will need to save it off and then create a new one
with the exact same IMGLIST command run from a box running SSEC McIDAS
(i.e., from a box where you get back no information).

Tom