Hi! I've been running a prefect server instance (v...
# ask-communityv
Vincent Chéry
08/25/2021, 12:14 PMHi! I've been running a prefect server instance (version 0.14.11) in production for a few months, running a dozen flows, some of them every minute. The UI is getting extremely slow when clicking on a flow run or task run. Is this known to happen ? Should I delete old flow and task runs to limit the size of flow_runs and task_runs tables ? If yes, what's the best way to do it ? Thanks in advance !
Vincent Chéry
08/25/2021, 12:50 PMI just checked in the database how many rows there are in both tables:
Copy code
prefect_server=# SELECT COUNT(*) FROM flow_run;
count
--------
169607
(1 row)
prefect_server=# SELECT COUNT(*) FROM task_run;
count
---------
1501563
(1 row)Vincent Chéry
08/25/2021, 12:51 PMThat's a lot but should not really be a problem with indexes.. any idea why it's getting so slow ?
k
Kevin Kho
08/25/2021, 2:01 PM
Hey @Vincent Chéry, maybe the database is just getting slow? Yes this is known to happen, server users regularly truncate the
task_runlogsv
Vincent Chéry
08/25/2021, 2:02 PMThanks Kevin
Vincent Chéry
08/25/2021, 2:03 PMI started deleting archived versions of my flows, I'm guessing that will the cascade delete task_run and logs tables as well ?
Vincent Chéry
08/25/2021, 2:04 PMDo you have a recommendation for how to automate this ? Like how to truncate everything that's older than one month ?
k
Kevin Kho
08/25/2021, 2:04 PM
No I don’t believe it will cascade, you can delete based on time though (older than a month), and it should all be aligned
v
Vincent Chéry
08/25/2021, 2:07 PMIs there any way to do this through the python or grpahql APIs ?
k
Kevin Kho
08/25/2021, 2:10 PM
The API no. If Python, manually through connecting to postgres. This thread might help you.
v
Vincent Chéry
08/25/2021, 2:11 PMPerfect, thanks 😉
👍 1
2 Views