are you running the psql with -c or -f?
add -o to put all in a file and -e to write
the last query, and with a tail to the log, you can see where is
the problem
On 2020-07-30 10:11, Olivier Leprêtre
wrote:
Hi, thanks for your answers, Application_name is a good tip, 64 chars are enough to code steps, I'll use that I expected being able to write raise events to a table or to store data in another table but it doesn't appear to be possible due to transaction isolation. Thanks a lot. Olivier -----Message d'origine----- De : Julien Rouhaud <rjuju123@xxxxxxxxx> Envoyé : jeudi 30 juillet 2020 12:19 À : Adrian Klaver <adrian.klaver@xxxxxxxxxxx> Cc : Olivier Leprêtre <o.lepretre@xxxxxxxxx>; pgsql-general <pgsql-general@xxxxxxxxxxxxxx> Objet : Re: Track pgsql steps On Wed, Jul 29, 2020 at 7:58 PM Adrian Klaver <adrian.klaver@xxxxxxxxxxx> wrote:On 7/29/20 8:44 AM, Olivier Leprêtre wrote:Hi, I have a rather long pgsql procedure and I would like to detect which step is currently executing (subscript 1,2,3…). Due to transaction isolation, it’s not possible to make it write in a table or get nexval from a sequence because values become available only after the complete end of the procedure. Do you see any solution in this purpose ?RAISE NOTICE?: https://www.postgresql.org/docs/12/plpgsql-errors-and-messages.html#PL PGSQL-STATEMENTS-RAISEYou can also abuse SET application_name, as the value will be directly seen by other transactions. You're quite limited in the number of bytes to use, but if you just need to do some quick testing it can be helpful.
