The Geotrellis openEO back-end supports passing additional options at job creation time to tweak batch jobs:
connection.create_job(process_graph, additional={
"driver-memory": "8G",
"driver-memoryOverhead": "2G"
})
data_cube.create_job(job_options={
"sentinel-hub": {
"client-alias": "vito",
"input": "sync"
}
})
To reduce:
- the amount of irrelevant logs for users to wade through,
- the amount of storage for these logs on disk,
the openEO API spec provides a log_level
to set at job creation time,
for example with create_job()
or execute_batch()
:
cube.create_job(
log_level="warning",
)
(Note that, originally, this log level could also be set through a "logging-threshold" job option, but that usage pattern is deprecated in favor of the standardized way mentioned above)
This log level effectively prevents log entries with a lower level (in this case: "debug" and "info") from being written to the logs.
It accepts the log levels as defined in the OpenEO API: "debug", "info", "warning" and "error" and defaults to "info".
Note: logs originating from openeo.udf.debug.inspect
are unaffected by this threshold; regardless of their log
level, be it "debug", "info", "warning" or "error", they will always be written to the logs.
Setting the "allow_empty_cubes" flag (defaults to: false
) enables working with empty data cubes
where this otherwise would raise an error. It is applied to all collections within this job.
job_options={
"allow_empty_cubes": True
}