Обсуждение: Re: pg_autovacuum does not start on system startup - 8.0 Release -

> -----Original Message-----
> From: Harald Massa [mailto:ghum@gmx.net]
> Sent: 24 January 2005 10:45
> To: Dave Page
> Cc: pgsql-hackers-win32@postgresql.org
> Subject: Re: [pgsql-hackers-win32] pg_autovacuum does not
> start on system startup - 8.0 Release -
>
> Dave,
>
> > > As said... I can start that service manually after Windows
> > > starts talking to
> > > me. What can I do to make it run properly?
> >
> > Hmm, I'll have to think about the best way to do that.
>
> My impression is that Service-startup is rather
> nondeterministic in aspects
> of timing.
>
> So I would suggest to integrate that sleepy workaround within
> pg_autovacuum
> service. Just let it sleep for 'some time' ... maybe even
> configurable ...
> before it connects. (recommended default: 180 seconds; NOT
> microseconds and
> NOT milliseconds :))

Uh-huh :-)

The problem with that is that with service start taking 3 minutes,
restarting the service manually could be somewhat annoying for simple
config changes (bear in mind that the service doesn't report that it is
running until it has connected successfully to the database). What about
a command line option to specify a delay? That way the user will know
about it because they will have specified it?

Regards, Dave.

Dave,

> > So I would suggest to integrate that sleepy workaround within
> > pg_autovacuum service. Just let it sleep for 'some time' ... maybe even
> > configurable before it connects.
> Uh-huh :-)
>
> The problem with that is that with service start taking 3 minutes,
> restarting the service manually could be somewhat annoying for simple
> config changes (bear in mind that the service doesn't report that it is
> running until it has connected successfully to the database). What about
> a command line option to specify a delay? That way the user will know
> about it because they will have specified it?

Yeah, I meant that when writing "configurable" ... that is ment to be a
command line parameter. So maybe the default should be 0 to 3 seconds.
(and the command line switch should expect seconds)

Harald

Dave Page wrote:

>The problem with that is that with service start taking 3 minutes,
>restarting the service manually could be somewhat annoying for simple
>config changes (bear in mind that the service doesn't report that it is
>running until it has connected successfully to the database). What about
>a command line option to specify a delay? That way the user will know
>about it because they will have specified it?
>

One answer might be to  have pg_autovacuum not be so pessimistic failed
connections.  I wrote it so that pg_autovacuum exits if it fails to
connect.  Perhaps instead, when pg_autovacuum fails to connect it should
sleep for a while and then retry.   Perhaps the best plan would be a new
command line option that when used tells pg_autovacuum to continue to
retry for ever, when not used, it will still work the way it always did.

Matthew