[gradsusr] GrADS vs GDS data discrepancies

[Matt Alonso]

Hi Jennifer,

Ugh, not sure how I missed that.  I assumed there was a user error
somewhere in there :(

2.0.2 has been a very stable workhorse of mine for several years - thanks
to you and Brian!  So there was never a reason to change.  While attempting
to debug prior to emailing, I did try using
grads-2.1.0-bin-CentOS5.11-x86_64.tar.gz
but, as you mention in the changelog for 2.1.1.b0, there was an issue w/the
ssl libs.  I did get 2.1.1.b0 working though.  One thing I did just notice
though is that plotting the same metric from a local netcdf (8007x4004)
takes about twice as long using v2.1.1.b0 (30 seconds) compared to 2.0.2
(14 seconds).  That is certainly not a complaint in any way, shape, or
form, but this node generates 1,000's of maps a day, so this would make a
material difference to load and performance.

After using the patched subset.gs you provided, everything is working
perfectly in 2.1.1.b0!  The seg fault is still there in 2.0.2 though.  I
will dig into that a little bit more.

Thanks again!

Regards,
Matt


On Tue, Feb 14, 2017 at 10:21 AM, Jennifer M Adams <jadams21 at gmu.edu> wrote:

> Matt,
> The problem with undef values is likely related to this note on the GDS
> home page:
>
> "The release of GrADS version 2.0.a6 implemented some necessary changes
> related to the handling of undefined values. These changes can affect the
> GDS. If you are administering a GDS version 2.0 with GrADS 2.0.a6 or higher
> behind it, then a 1-line patch is needed for the file $GDSHOME/scripts/
> subset.gs. The patched version is here
> <http://cola.gmu.edu/grads/gds/patched_subset.gs>. Updating the subset.gs
> script does not require a reinstallation or even a reload of your GDS."
>
> You should be using a newer version of GrADS anyway … why 2.0.2 ???
>
> As for the seg fault, it’s impossible to diagnose without knowing more
> about your data file. Try comparing the ncdump output on your native file
> and the GDS URL and see where they differ.
> —Jennifer
>
> On Feb 11, 2017, at 3:18 PM, Matt Alonso <matt.alonso at GMAIL.COM> wrote:
>
> Hi all,
>
> Long time user of GrADS but first time user of GDS.  When attempting to
> plot the same dataset locally vs going through GDS, I am running into some
> strange behavior.
>
> Background:
> - same CentOS 6 machine is being used to run GrADS and GDS
> - GrADS v2.0.2 binaries
> - GDS 2.0
> - data is netcdf but I am using a ctl file for templating purposes
> - dataset entry in GDS xml file:
>
>       <dataset name = "grid_hist"
>                file = "/path/to/data.ctl"
>                url = "http://www.iges.org/grads/gds/"
>            format = "nc">
>       </dataset>
>
> - the ctl file in the dataset path (above) is (literally) the exact same
> path used when I mention the "local dataset"
> - each netcdf file is ~65MB
>
> Two issues:
>
> 1)
>
> When opening the data via GDS, GrADS does not appear to identify
> undefined/missing values.  The undef value in the ctl and netcdf file is
> -999.0.  When using "set gxout stat" on the local dataset for an arbitrary
> domain and variable, it shows an Undef count > 0 and a Min of 0, but the
> GDS file shows Undef count = 0 and a Min = Undef value.
>
> 2)
>
> When I open either the GDS or local data set, the lon/lat extent printed
> is:
>
> LON set to -180 179.945
> LAT set to -90 89.9725
>
> If I do not change any of the dimensions and try to display any metric in
> the file - using any gxout - the local dataset opens fine, however, using
> the data in GDS causes GrADS to segfault.  There are no errors in gds.log,
> it even shows "success" in the FINISHED record.
>
> If I DO change the lat/lon to be any subset of the entire domain, then
> both the local and GDS datasets plot fine - ignoring the behavior in #1
> above.
>
> I am sure I am doing something wrong but do not seem to be able to figure
> out what.  Thanks in advance for your help!
>
> Regards,
> Matt
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr
>
>
> --
> Jennifer Miletta Adams
> Center for Ocean-Land-Atmosphere Studies (COLA)
> George Mason University
>
>
>
>
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://gradsusr.org/pipermail/gradsusr/attachments/20170215/35f94f80/attachment.html