I found out in [1] that I am not correctly tracking the workers for vacuum commands. I trap workers used by parallel_vacuum_cleanup_all_indexes but not parallel_vacuum_bulkdel_all_indexes.

Back to the drawing board.

[1] https://www.postgresql.org/message-id/flat/783bc7f7-659a-42fa-99dd-ee0565644...@dalibo.com

--
Benoit Lobréau
Consultant
http://dalibo.com


Reply via email to