Обсуждение: Re: [HACKERS] statement_timeout logging

pgman wrote:
> Simon Riggs wrote:
> > Currently, when we set a statement_timeout and a query runs over that
> > time there is no log message to say that the statement has timed out. We
> > do get a message which says
> >     ERROR:  canceling query due to user request
> > and so in the server log it is impossible to tell the difference between
> > a query that has been cancelled deliberately by the user and a query
> > whose time(out) has come.
>
> I have updated the message to:
>
>      errmsg("canceling query due to user request or statement timeout")));

Oops.  Did we freeze the message strings already for this release?

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073

On Fri, Sep 16, 2005 at 03:41:11PM -0400, Bruce Momjian wrote:
> pgman wrote:

> > I have updated the message to:
> >
> >      errmsg("canceling query due to user request or statement timeout")));
>
> Oops.  Did we freeze the message strings already for this release?

Not yet.

I'm not sure I agree with the wording though -- is there no way to tell
the difference?

--
Alvaro Herrera -- Valdivia, Chile         Architect, www.EnterpriseDB.com
"No single strategy is always right (Unless the boss says so)"
                                                  (Larry Wall)

>Alvaro Herrera
> On Fri, Sep 16, 2005 at 03:41:11PM -0400, Bruce Momjian wrote:
> > > I have updated the message to:
> > >
> > >      errmsg("canceling query due to user request or
> statement timeout")));
> >
> > Oops.  Did we freeze the message strings already for this release?
>
> Not yet.
>
> I'm not sure I agree with the wording though -- is there no
> way to tell
> the difference?

Yes, my patch. :-)

The message is generated following receipt of a SIGINT which is sent by
kill() after the statement_timeout.

My patch was able to generate a different message by logging just before
the kill() is called. (Obviously, there isn't much "after" in this
circumstance).

Best Regards, Simon Riggs

Simon Riggs wrote:
> >Alvaro Herrera
> > On Fri, Sep 16, 2005 at 03:41:11PM -0400, Bruce Momjian wrote:
> > > > I have updated the message to:
> > > >
> > > >      errmsg("canceling query due to user request or
> > statement timeout")));
> > >
> > > Oops.  Did we freeze the message strings already for this release?
> >
> > Not yet.
> >
> > I'm not sure I agree with the wording though -- is there no
> > way to tell
> > the difference?
>
> Yes, my patch. :-)
>
> The message is generated following receipt of a SIGINT which is sent by
> kill() after the statement_timeout.
>
> My patch was able to generate a different message by logging just before
> the kill() is called. (Obviously, there isn't much "after" in this
> circumstance).

Your patch just prints a statement "before" printing the cancel message.
What we should do is to print the proper cause of the cancel, and I am
working on a patch for that.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073