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
When monitoring the postgresql log, it has been observed that the loop inside the main, static background worker process that starts a dynamic background for each database using pg_partman does not appear to pause between dynamic bgw registrations. This causes excessive system load and can also cause postgresql log files to fill up extremely fast depending on logging levels.
This problem does not exist in 9.4 or 9.5
As a workaround for now, the bgw can be disabled in postgresql.conf by removing the entry from the shared_preload_libraries option and scheduling the run_maintenance() plpgsql function to run via an external process (cron, etc).
The text was updated successfully, but these errors were encountered:
For anyone else that reads this, current status and discussion are under the below thread on the bugs list. Title of the thread is misleading as that was my initial diagnosis of the problem. The problem has now been narrowed down to a change in the Latch handling code in 9.6
When monitoring the postgresql log, it has been observed that the loop inside the main, static background worker process that starts a dynamic background for each database using pg_partman does not appear to pause between dynamic bgw registrations. This causes excessive system load and can also cause postgresql log files to fill up extremely fast depending on logging levels.
This problem does not exist in 9.4 or 9.5
As a workaround for now, the bgw can be disabled in postgresql.conf by removing the entry from the
shared_preload_libraries
option and scheduling therun_maintenance()
plpgsql function to run via an external process (cron, etc).The text was updated successfully, but these errors were encountered: