Gpt hangs during polarimetry graph


#1

Hi,
I’m running a single graph of polarimetric decomposition on Sentinel-1 scene using gpt.

The command hangs once the processing is part complete (1/3 judging by this image) with no error message.

I am using SNAP 5 and have allocated around 24GB RAM to gpt.
Any help would be appreciated! For now I will test a different scene.


#2

Hi @sandersson

If the process has been killed, you should see some message from gpt. Can you try again and post a screenshot of GPT?

M


#3

Should yes, but in the end it just hangs and it doesn’t say why it stops processing.
Thanks,
Simon


GRD-Border-Noise and Thermal noise removal are not working anymore since March 13. 2018
#4

I just see that it is using restituted orbits instead of precise orbits. Update SNAP to version 6 and try to run the graph again, let’s see.

M


#5

Updated to SNAP 6.
Graph gets to 20% and says ‘Database closed’ and then hangs.

No error message. My intuition says it’s a resource issue but there seems to be enough space and RAM, plus the graph works in the GUI.

Frustrating because I am running a service based on automated SNAP processing with gpt, and I think this is a bug created by the Sentinel-1 format change!

See also:



#6

The steps in the graph are:

  1. Read
  2. Apply-Orbit-File
  3. TOPSAR-Deburst
  4. Polarimetric-Matrices
  5. Polarimetric-Speckle-Filter
  6. Polarimetric-Decomposition
  7. Terrain-Correction
  8. Write

Again the graph works in the SNAP GUI but not in the command line. Perhaps someone could test the graph ?

When I use “gpt -e” the graph hangs at just past 20% processing.

Regards,


#7

Resolved by calling gpt with the -c parameter set at 75% of the java heap space.

eg:
“gpt -c 24G”

The java heap space is set using -Xmx in gpt.vmoptions.
This has also reduced the time for my processing chain to run by a factor of 4

gpt --diag was also helpful.