Re: [PATCH] pg_isready (was: [WIP] pg_ping utility)

<p>On Jan 26, 2013 6:56 PM, "Craig Ringer" <<a href="mailto:craig@2ndquadrant.com">craig@2ndquadrant.com</a>>
wrote:<br/> ><br /> > On 01/27/2013 06:20 AM, Phil Sorber wrote:<br /> > > On Sat, Jan 26, 2013 at 4:37 PM,
PavelStehule <<a href="mailto:pavel.stehule@gmail.com">pavel.stehule@gmail.com</a>> wrote:<br /> > >>
2013/1/26Phil Sorber <<a href="mailto:phil@omniti.com">phil@omniti.com</a>>:<br /> > >>> On Sat, Jan
26,2013 at 12:39 PM, Pavel Stehule <<a href="mailto:pavel.stehule@gmail.com">pavel.stehule@gmail.com</a>>
wrote:<br/> > >>>> 2013/1/26 Phil Sorber <<a
href="mailto:phil@omniti.com">phil@omniti.com</a>>:<br/> > >>>>> On Sat, Jan 26, 2013 at 11:53 AM,
PavelStehule <<a href="mailto:pavel.stehule@gmail.com">pavel.stehule@gmail.com</a>> wrote:<br /> >
>>>>>>2013/1/26 Phil Sorber <<a href="mailto:phil@omniti.com">phil@omniti.com</a>>:<br /> >
>>>>>>>On Sat, Jan 26, 2013 at 4:02 AM, Pavel Stehule <<a
href="mailto:pavel.stehule@gmail.com">pavel.stehule@gmail.com</a>>wrote:<br /> > >>>>>>>>
Hello<br/> > >>>>>>>><br /> > >>>>>>>> We now haw to solve small
puppetissue, because our puppets try to<br /> > >>>>>>>> start server too early, when old
instancelive still.<br /> > >>>>>>>><br /> > >>>>>>>> Maybe some
newparameter - is_done can be useful.<br /> > >>>>>>>><br /> >
>>>>>>>What about something like:<br /> > >>>>>>> pg_isready; while [ $?
-ne2 ]; do sleep 1; pg_isready; done<br /> > >>>>>> it is not enough - server is done in a moment,
wherecan be started<br /> > >>>>>> again - or when we can do safe copy of database data
directory.<br/> > >>>>>><br /> > >>>>> I guess i am not completely understanding
thecase you are trying to<br /> > >>>>> solve. Can you explain a bit further?<br /> >
>>>>We use puppets and due some simplification we cannot to use reload<br /> > >>>> when
configurationis changed. Our puppets has not enough intelligence<br /> > >>>> to understand when is
reloadenough and when is restart necessary. So<br /> > >>>> any change to configuration require
restartingpostgres. I don't know<br /> > >>>> why "service restart" are not used. I believe so our
puppetguru know<br /> > >>>> it. It just do sequence STOP:START  Now puppets are "smart" and able<br />
>>>>> to wait for time, when server is ready. But there are missing simple<br /> > >>>>
testif server is really done and I see a error messages related to<br /> > >>>> too early try to start.
Sosome important feature can be verification<br /> > >>>> so server is really done.<br /> >
>>>><br/> > >>>> We can do it with test on pid file now - and probably we will use it.<br />
>>>>> But I see so this is similar use case (in opposite direction)<br /> > >>>><br />
>>>> I guess I am still not clear why you can't do:<br /> > >>><br /> > >>>
stop_pg_via_puppet<br/> > >>> pg_isready<br /> > >>> while [ $? -ne 2 ]<br /> > >>>
 do<br /> > >>>     sleep 1<br /> > >>>     pg_isready<br /> > >>>   done<br />
>>>> do_post_stop_things<br /> > >>> start_pg_via_puppet<br /> > >>><br /> >
>>because ! pg_isready <> pg_isdone<br /> > >><br /> > > So you are proposing a different
utility?Sorry, I thought you were<br /> > > proposing a new option to pg_isready. What would pg_isdone be
testing<br/> > > for specifically? Is this something that would block until it has<br /> > > confirmed a
shutdown?<br/> ><br /> > That's what it sounds like - confirming that PostgreSQL is really fully<br /> > shut
down.<br/> ><br /> > I'm not sure how you could do that over a protocol connection, myself.<br /> > I'd just
readthe postmaster pid from the pidfile on disk and then `kill<br /> > -0` it in a delay loop until the `kill`
commandreturns failure. This<br /> > could be a useful convenience utility but I'm not convinced it should be<br />
>added to pg_isready because it requires local and possibly privileged<br /> > execution, unlike pg_isready's
networkbased operation. Privileges could<br /> > be avoided by using an aliveness test other than `kill -0`, but
you<br/> > absolutely have to be local to verify that the postmaster has fully<br /> > terminated - and it
wouldn'tmake sense for a non-local process to care<br /> > about this anyway.<br /> ><p>Maybe something to add to
pg_ctl?<p>>--<br /> >  Craig Ringer                   <a
href="http://www.2ndQuadrant.com/">http://www.2ndQuadrant.com/</a><br/> >  PostgreSQL Development, 24x7 Support,
Training& Services<br /> ><br />