iyashk-DB
Databricks Employee
Databricks Employee

VACUUM works fine on DLT tables running in continuous mode. DLT does automatic maintenance (OPTIMIZE + VACUUM) roughly every 24 hours if the pipeline has a maintenance cluster configured.

Q: The liquid cluster is enabled in the tables. Will Vacuum work on these tables when it is getting loaded in continuous mode? How to run the vacuum without impacting the checkpoints and the actual load of the DLT tables?
A: It won’t mess with checkpoints: VACUUM only removes orphaned data files and skips special dirs like _delta_log; DLT manages streaming checkpoints separately under the pipeline storage .../checkpoints/. Keep retention ≥ 7 days to stay safe.

View solution in original post