Thank you – we
really appreciate your tracking down the issue.
We tested the
Hovmoller plots and animation for GHRSST NCDC AVHRR_OI on
our test server, which is running LAS 7.3 and Ferret 6.842.
On our test server, instead of getting the error we reported
earlier in this thread, we get a pink stripe on a red
background. (see attached) Our public operational server
is LAS 7.3 and Ferret 6.72, and has the “remote server
unable to deliver data” (as reported earlier).
Then we looked
at the NODC LAS server, which is using LAS 7.3, Ferret
6.82. The same hovmoller plot returned a result with the
pink stripe, red background, just like our test server.
This seems to
indicate that the older version of Ferret handles the NetCDF
library error in a different way (giving an explicit error
return) than the more recent Ferret (strange plot
behavior difference corroborate the NetCDF library error
please outline what the tests were and how the tests
accessed our data – ie, using our THREDDS URL link to a
single variable and time index, or downloading files via FTP
and running the decompressed files? We are trying to build
up a suite of tests to help our diagnostics, and confirm the
data integrity in our cache.
currently trying to get LAS 8 set up in a test environment –
if a solution is found for this issue for LAS 8, we will
step up our timeline to switch to LAS 8.
Thank you very
Both this issue with animations and your question last week
about Hovmuller plots with datasets having long time axes are
running into an error in the NetCDF libraries when we try to
read strided (decimated) grids of data.
I've tested these kinds of read operations on your datasets
with with the various versions of NetCDF, and none of them
return a correct result. I have made a report to the authors
of the NetCDF library.
I see that your LAS is currently running version 7.3. When we
do have a fix for this, it will require an upgrade to your
LAS, to the newer LAS Version 8.
On 5/1/2013 10:17 AM, Ansley Manke wrote:
This is very likely the same issue that you wrote about last
week. Again, we're asking for a portion of the data, in
this case a number of time steps spaced across the time
region you have asked for. The first step of making an
animation is to get the range of the data represented, so
that the color scale can be the same across all the frames
of the animation. Once that's done, we can just read one
timestep at a time and make the image.
I'll try to get back to you in the next couple of days.
On 4/30/2013 2:47 PM, Shelton, Kacie E
We are having issues with LAS
animation, and it nominally appears to be the same or
similar error with our datasets that are very large
(either volume or time series).
The behavior: we can only request
animation for a very small area for only a few days. If
an error returns, we need to close the tab, return to the
main LAS window, and click on Animation again – basically
trial and error to find how many granules it will animate
in one request. 6 granules is the limit for the dataset
GHRSST AVHRR_AMSR_OI. The error is returned before the
“Stop Downloading” button appears.
Cached data cleared from
Malformed or inaccessible DAP DATADDS (OPeNDAP/netCDF
Error code -73)
I am attaching an image of the error,
and a rtf of the long error report.
Any idea what is going on?