Re: LISTEN NOTIFY sometimes huge delay

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: "Peter Eser HEUFT [Germany]" <peter(dot)eser(at)heuft(dot)com>
Cc: pgsql-performance(at)lists(dot)postgresql(dot)org
Subject: Re: LISTEN NOTIFY sometimes huge delay
Date: 2022-04-28 16:38:35
Message-ID: 21909.1651163915@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

"Peter Eser HEUFT [Germany]" <peter(dot)eser(at)heuft(dot)com> writes:
> I have a table with time series data and on this table a trigger for
> notifies:
> containers_notify AFTER INSERT ON containers FOR EACH ROW EXECUTE
> PROCEDURE containers_notify('containers_notify_collector')
> and the function does:
> PERFORM pg_notify(CAST(TG_ARGV[0] AS text), row_to_json(NEW)::text);

> This works as a charm but occasionally (I think with more load on the
> system) the notifications are received much time (up to hours!) after
> the INSERTs.
> Nevertheless no notifications become lost, they are only very late! The
> delay grows, seems as a queue grows, but the java process tries to fetch
> the notifications fairly fast,

Hm. We've not previously had reports of late notifications. One idea
that comes to mind is that the server won't deliver notifications as
long as the client has an open transaction, so is it possible your
listening process sometimes forgets to close its transaction?

> Versions:
> PostgreSQL 10.12 on x86_64-pc-linux-gnu, compiled by
> x86_64-pc-linux-gnu-gcc (Gentoo 6.4.0-r1 p1.3) 6.4.0, 64-bit
> JDBC 42.2.23

That's pretty old. We've made a number of changes to the LISTEN/NOTIFY
code since then; although in reading the commit log entries about them,
nothing is said about long-delayed notifications.

regards, tom lane

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message aditya desai 2022-04-29 10:59:22 FATAL: canceling authentication due to timeout
Previous Message Peter Eser HEUFT [Germany] 2022-04-28 14:28:15 LISTEN NOTIFY sometimes huge delay