if a graph has a "high frequency" tick_recurring node (which won't be the case in most production cases), the historical computation records for the completed ticks (and, perhaps, its triggered downstream nodes) will start adding up after a while.
provide the ability to compact completed tick_recurring record, to save storage / bandwidth / memory.
example:
in the jourdash delivery app, a couple of "deliveries" waited for pickup or dropoff for weeks, while generating frequent (:time_simulation) ticks and processing them as (:driver_location_current_update) simulated gps updates. after a while, that added up to a lot of records. these records serve as historical records, but (at least in this simulated case) might add up to a lot of records, and it would be nice to have the option to have them automatically compressed away.
=> select state, node_name, computation_type, count(*) from computations where execution_id='EXEC01T7998GRG8LD7AGB9R0' group by state, node_name, computation_type;
state | node_name | computation_type | count
-----------+--------------------------------+------------------+-------
success | rating_reminder | compute | 1
abandoned | trip_history | compute | 1
success | payment_collection | compute | 1
abandoned | rating_reminder | compute | 1
success | created_at | compute | 1
abandoned | driver_location_current_update | mutate | 27
success | rating_reminder_timer | tick_once | 1
success | driver_location_current_update | mutate | 24059
success | trip_history | compute | 9
abandoned | time_simulation | tick_recurring | 31
success | time_simulation | tick_recurring | 24084
success | current_activity | compute | 12
success | trip_completed_at | compute | 1
not_set | time_simulation | tick_recurring | 1
(14 rows)
if a graph has a "high frequency"
tick_recurringnode (which won't be the case in most production cases), the historical computation records for the completed ticks (and, perhaps, its triggered downstream nodes) will start adding up after a while.provide the ability to compact completed tick_recurring record, to save storage / bandwidth / memory.
example:
in the jourdash delivery app, a couple of "deliveries" waited for pickup or dropoff for weeks, while generating frequent (
:time_simulation) ticks and processing them as (:driver_location_current_update) simulated gps updates. after a while, that added up to a lot of records. these records serve as historical records, but (at least in this simulated case) might add up to a lot of records, and it would be nice to have the option to have them automatically compressed away.