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

20040105: Unidata McIDAS-X Version 2003 at NIU



>From: Gilbert Sebenste <address@hidden>
>Organization: NIU
>Keywords: 200401052323.i05NNbp2019777 McIDAS-X ldm-mcidas

Hi Gilbert,

>In response to the new products on McIDAS, I did what you said to do in 
>your email to add the new products.

OK.  Presumably you mean you added the new products to the McIDAS
routing table?

>mainly, to download and run the new 
>batch files to get and decode the new CIMMS and other new images.

OK.

>But...
>When I did the first of these two commands:
>
>>   batch.k XCD.BAT
>>   batch.k XCDDEC.BAT
>>
>> - adjust the entries in ~ldm/decoders/mcscour.sh; added the scouring of
>>   the new MD files produced for ETA and GFS MOS output
>
>then it bombs with "string not found: XCDDATA"

Why did you run these two commands?  These setup XCD decoding of LDM
ingested conventional data.  They have nothing to do with the decoding
of Unidata-Wisconsin (IDD UNIWISC aka MCIDAS) data.

>Hmmm. What am I doing wrong?

When setting up XCD decoding, you have to define XCDDATA as the
directory where the XCD decoders create their output.  For instance, if
you were decoding into /data/ldm/mcidas, then you would do the
following in the 'mcidas' account to setup XCD decoding:

<login as 'mcidas'>
cd workdata
cp ~mcidas/data/SCHEMA /data/ldm/mcidas
cp ~mcidas/data/SYSKEY.TAB /data/ldm/mcidas
cp ROUTE.SYS /data/ldm/mcidas
te.k XCDDATA \"/data/ldm/mcidas
batch.k XCD.BAT
batch.k XCDDEC.BAT

>The ADDE server seems to think that everything is cool. I tried it out...

OK, sounds good.

>From: Gilbert Sebenste <address@hidden>
>Date: Mon, 5 Jan 2004 17:24:35 -0600 (CST)

>Tom,
>
>BTW...
>
>> - created the user mcadde and put it in group unidata
>> 
>> - downloaded, built, and installed McIDAS-X v2003; suprisingly, this
>>   was not even close to a speed record (6:30).  The speed record
>>   is held by a single processor AMD 2000+ box at the Universidade
>>   Federal do Rio de Janeiro (5:37).  I think that the limiting
>>   factor is the disk subsystem.  The machine in Rio has ATA 133
>>   hard disks; what does weather3 have?
>
>an IDE hard drive. Yeah, I couldn't afford SCSI for the grant. That's what 
>slowed it down.

Yes, but the machine in Rio doesn't have SCSI disks either.  As an
update, the same machine in Rio was upgraded to an AMD 2800+ processor,
and its system disk (as opposed to RAID disks) was upgraded from a 40
GB ATA-100 to an 80 GB ATA-133.  I built McIDAS on the new
configuration, and that build took all of 5:07.  Yet another new speed
record for that box!

As a comparison, I did the same build on a dual AMD 2600+ machine that
has an ATA-133 80 GB system disk here at the UPC, and that build took
around 8 minutes, but the system was ingesting all IDD feeds as part of
its stress testing.

Tom