Re: Reuse of REF Cursor

On 4/11/21 1:02 PM, Abraham, Danny wrote:
> 2021-04-09 08:00:08.692 IDTERROR: canceling statement due to statement timeout
> 2021-04-09 08:00:08.692 IDTCONTEXT: PL/pgSQL function orhpans_active_clean_table(character varying,integer) line 42 at FETCH
> PL/pgSQL function orhpans_active_removal() line 31 at assignment
> PL/pgSQL function ajf_backup(integer) line 39 at assignment
>
> Can a FETCH fail if the table is locked? The FETCH is stuck for the <statement_timeout> time.
>
> Should I lock all tables involved with the query?
>
> Any specific time-out on the fetch? Or should I use the general statement-timeout?
>
> I mean move from regular programming mode to paranoidic mode....
>
> The failure is inconsistent.. Never fails in PG 11.5, but fails in PG9.5.5 about once a week...
>
> I need a full understanding of the problem in order to force big,slow customers to migrate to PG11.5.

9.5.21 would be an important step.  Heck, it might solve the problem.

--
Angular momentum makes the world go 'round.