error in getting gfs-data

[df8oe]

i installed the predictor today and I think it is working - exept it gets no gfs-data. The progress tells "Server says: downloaded 0% of GFS files" for 20 times followed by the message"There was an error in running the prediction: Unknown error exit".

The directory "gfs" exists and is writeable, and I correct the path in predict.py. Temporary folder /tmp/pydap-cache was created by the program - but it is still empty (so the same with the pfs - directory).

In py.log stands

INFO: Looking for latest dataset which covers .....
INFO: Picked IP: 140.90.33.61
ERROR: Could not locate a dataset for the requested time

The clock is running on local time (Europe/Berlin) - ok, I must fix the error that a time between UTC and MESZ throws a "...is in the past error". But when I enter a time 5 hours in the future (or a day), py.log - error remains the same...

Is there a possibility to get verbose error messages??

Is it possible to add variables to the sites.json for burst altitude, ascent/descent rate for each location?

Andreas

[danielrichman]

Hi Andreas,

We're getting that error too today - the NOAA servers are running slowly / on the blip.

You can't add variables to sites.json for balloon settings, only launch location. You could edit index.php to change the defaults.

[df8oe]

Many thanks for ypur fast reply. This is the problem - but it persists today, too. May be noaa-servers are overloaded e.g.

I tried another balloon-forecast-site http://weather.uwyo.edu/polar/balloon_traj.html and it works - habhub has problems in getting gfs-data. And today in the morning I recognized that the new model 00z was available there 30 minutes before the model arrived at habhub.... Are there mirrors or other gfs-servers which are quicker in getting new models and/or have more bandwidth and/or better stability? I see the used server can be modified in predict.py...

Andreas