Re: [thredds] Questions concerning TIME joinExisting aggregations

Hi Mike:

On 1/4/2011 5:01 PM, McCann, Mike wrote:
Hello,

I am testing some aggregations of real-time /in situ /oceanographic mooring data served by the THREDDS Data Server.

Generally, the TDS has been working pretty well. I do, however, see some inconsistent behavior that bay be the result of how I've built the catalog or may be issues with the THREDDS Data Server. I'd appreciate any information that will help with these two issues:

   1. The .html form does not properly show the indices of the
      variable's coordinate axes.  For instance, for the data set
      
<_http://elvis.shore.mbari.org/thredds/dodsC/moorings/OS_MBARI-M1_TS.html_>
      clicking on the TEMP variable shows "0:1:0" for each of the
      time, depth, latitude, and longitude coordinates. The form seems
      to know how many points are in the coordinates as when you
      select them on the form they are properly shown. Shouldn't the
      variable's coordinates show the same index limits that are shown
      for the coordinate variables?

Ill see if we can get that fixed.

   1. I was getting errors for an unconstrained requests of TEMP the
      M0 & M1 moorings, but not for the M2 mooring, e.g.:


    _Mooring M1:
    
_http://elvis.shore.mbari.org/thredds/dodsC/moorings/OS_MBARI-M1_TS.ascii?TEMP

        Returned: NcSDArray java.lang.ArrayIndexOutOfBoundsException
        =null for request= NcSDArray read TEMP TIME(0,1,52569)
        DEPTH(0,1,10) LATITUDE(0,1,0) LONGITUDE(0,1,0) dataset=
        moorings/OS_MBARI-M1_TS


    _Mooring M2:
    
_http://elvis.shore.mbari.org/thredds/dodsC/moorings/OS_MBARI-M2_TS.ascii?TEMP

        Returned data



    So, guessing that this problem might have something to do with the
    THREDDS cache I removed the files in content/thredds/cache/agg.
    The problem went away!

    It seems that I should probably be using something like the
    recheckEvery attribute in my aggregation, but the documentation at
    <http://www.unidata.ucar.edu/software/netcdf/ncml/v2.2/Aggregation.html>
    seems to imply that it works for only aggregations created by the
    scan element.  I am using the netcdf element to stitch together
    several files with the last one updating about every 2 hours. If I
    set recheckEvery = "2 hours" will that solve my problem?


Can you send me your config catalog / ncml aggregation for the good and the bad datasets?

Do the updating files append only or get completely rewritten ?

I think recheckEvery works only with scans.

John