SpaceEye; preprocess.s1.preprocess task fails

[Aca-Ali]

While running the SpaceEye workflow, the algorithm fails at the S1 preprocessing task. This happens when I try to run the algorithm for a longer period of time (more than approximately 10 days). Attached you can find the run monitor and also the error description screen grab.
Also, I was wondering if there is a way to clear the cache?

PS. here is the VM specs I am using: Ubuntu 20.04, 4 CPUs, 17 GB memory, and around 600GB storage.

Thanks

[Renato L. de F. Cunha]

Hello @Aca-Ali. From the exception text you pasted above, the subprocess FarmVibes.AI creates to run the preprocess.s1.preprocess operation is terminating abnormally.

What that means is that something is killing the process. Usually, that happens due to the Operating System's Out-Of-Memory killer being activated because you don't have enough RAM to run the op.

You could might try these options:

• Reducing the number of workers with kubectl scale deployment terravibes-worker --replicas=3
• Running the spaceeye interpolation workflow
• Running FarmVibes.AI in a machine with more RAM

[Aca-Ali]

Thank you @renatolfc, I did not try the first two suggested solutions and used a machine with more RAM (the third solution) and it solved the issue.

[shenoy10]

Regarding the first solution provided here. What's the default number of replicas used?

[Rafael Soares Padilha]

The default number of replicas is max(1, CPU_COUNT // 2 - 1), in which CPU_COUNT is the number of available cores in your machine.