[Help]: sbas.compute_align() freezes in low RAM environment

[georgeboldeanu]

Hello Alexey! I try to run some processing chain for a combined SBAS and PS analysis on a small AOI, more exactly 4 km2. I do this using the latest docker image, which has the following pygmtsar version 2024.8.30.post3. All of this is necesarry because I try to limit the resources used by the environment to 4GB of RAM, cause the processing chain will be implemented on a low resources environment for small areas processing.

I run the docker as following:
docker run -p 8085:8085 -m 6g -v mount_volume:/home/data/ --rm -it --entrypoint bash pechnikov/pygmtsar. The data that I use is a whole year of Sentinel 1 data which is located in the mounted_volume.

When I use sbas.compute_align() the process stales in that point, no matter the time that I leave it (maximum time 4 hours), but when i hit stop from the jupyter lab notebook the process shows that somehow is completed, for the Aligning Repeat: step.

When I run the docker image without the memory constrictions everything runs ok, but the sbas.compute_align() takes too much RAM memory, so that is not a solutions (almost 30 GB of RAM are needed for the sbas.compute_align() function.)

Are there are known issues for this?!?

[AlexeyPechnikov]

You’ve allocated 6 GB of memory to the container using the -m 6g flag, but you haven’t specified a swap size.

It’s strange that aligning requires so much RAM. Are you processing individual bursts or entire subswaths?

[georgeboldeanu]

@AlexeyPechnikov Individual bursts.

[AlexeyPechnikov]

Try adding a couple of gigabytes of swap memory at first.