You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, we reserve 4GB of memory for the Delta Lake Java driver stack.
That used to be 1GB and then 2GB and now 4GB, as we dealt with the growing Observation table at BCH.
The hardcoded 4GB is overkill for sites (or even individual tasks) with smaller tables. And updating the hardcoded number for all cases whenever BCH hits a new limit seems bad.
So ideally the ETL would inspect the target table and we'd choose a memory size based on that. Nice and flexible for all cases.
The text was updated successfully, but these errors were encountered:
Currently, we reserve 4GB of memory for the Delta Lake Java driver stack.
That used to be 1GB and then 2GB and now 4GB, as we dealt with the growing Observation table at BCH.
The hardcoded 4GB is overkill for sites (or even individual tasks) with smaller tables. And updating the hardcoded number for all cases whenever BCH hits a new limit seems bad.
So ideally the ETL would inspect the target table and we'd choose a memory size based on that. Nice and flexible for all cases.
The text was updated successfully, but these errors were encountered: