ISSUE:
Evidently 0.7 memory drift over time to maximum of allocated memory (running in GCP kubernetes).
This occurs even if no new projects are created.
Snapshots (without datasets) are created every night.
Over a period of multiple days memory usage grows from say 3G to 5G (until whatever limit).
QUESTION:
How to configure evidently not to load all snapshot in memory at the same time and/or manage memory is a non divergent way -- to avoid eventual memory overflow?
ISSUE:
Evidently 0.7 memory drift over time to maximum of allocated memory (running in GCP kubernetes).
This occurs even if no new projects are created.
Snapshots (without datasets) are created every night.
Over a period of multiple days memory usage grows from say 3G to 5G (until whatever limit).
QUESTION:
How to configure evidently not to load all snapshot in memory at the same time and/or manage memory is a non divergent way -- to avoid eventual memory overflow?