Skip to main content
Version: Python

garbage_collect

The garbage_collect method runs full garbage collection. It is performed in Python first; then, the underlying JVM runs its garbage collector twice due to the cross-referencing nature of the Python/Java integration in Deephaven.

note

Since there is no way to force the Java garbage collector to run, the effect of calling this function is non-deterministic. Users should also be mindful of the overhead that running garbage collection generally incurs.

Syntax​

garbage_collect()

Parameters​

This method takes no arguments.

Example​

The following example uses Deephaven's perfmon (performance monitor) library's server_state method to print the amount of memory Deephaven is currently using at various points throughout the query - at the beginning, when objects are created and deleted, and finally after garbage_collect is called. The data in the UsedMemMiB column tracks how much memory is currently being used.

# Segment 1
from deephaven.perfmon import server_state
from deephaven import empty_table, garbage_collect
import gc, jpy

rt = jpy.get_type("java.lang.Runtime").getRuntime()
mem = server_state().view(["IntervalStart", "UsedMemMiB"]).reverse()
print(f"START {(rt.totalMemory()-rt.freeMemory())/1024/1024}")

x = [0]

# Segment 2
x[0] = empty_table(100_000_000).update("X = sqrt(i)")
print(f"CREATE {(rt.totalMemory()-rt.freeMemory())/1024/1024}")

# Segment 3
x[0] = empty_table(100_000_000).update("X = sqrt(i)")
print(f"CREATE {(rt.totalMemory()-rt.freeMemory())/1024/1024}")
del x
print(f"DELETE {(rt.totalMemory()-rt.freeMemory())/1024/1024}")
gc.collect()
garbage_collect()
print(f"GC {(rt.totalMemory()-rt.freeMemory())/1024/1024}")

img