Well Don, I'm responding to two things, one is that Dossy had asked about this some years back, and the other is the question that Agustin asked... all I've done so far is think about it and invite comment.
Having said that, could you look at the other change I made on my fork which is not going to be merged into the main repo unless someone looks at it and sees "yes, this does fix existing problems, and we should merge it"? There are other funcs where I think this should be performed, but, one at a time, and with review each time. On 2/25/12, Don Baccus <dhog...@pacifier.com> wrote: > > On Feb 25, 2012, at 1:45 AM, Jim wrote: > >> Hi Agustin, moving this to its own thread. >> >> On 2/25/12, Agustin Lopez <agustin.lo...@uv.es> wrote: >>> Hi all! >>> >>> Talking about nspostgres, has anybody implemented any way to auto >>> reconnect >>> to the database in a Postgresql start / stop? >>> >>> Thanks, >>> Agustin >> >> Since the normal procedure when restarting pg is to stop pg client >> programs, stop pg, start pg and start clients, I didn't see a use to >> try implementing this. > > That's right. > >> What I did see was extra code ran every time >> nspostgres was used to query (or use ddl/dml) to make sure a >> connection was stable, and that could be a heavy cost, especially if >> pg is not running on the same machine as its clients (incl aolserver). > > nspostgres implements recovery after the failure (crash) of a single > backend, not the starting/stopping of postmaster. > > That code's been around forever, is extremely stable, and while very few > people now use aolserver some very large sites depend on it remaining > stable. > > Unless you think you have a *really, really, really* good reason to muck > with it, please don't. > >> >> Having said that, I'd like to explore what would be required to make >> that happen. > > I don't see why. How often do you start/stop postmaster? If you're doing > so often, you're doing something wrong. Not in at least 2-3 years, so I'm presently thinking the same. They're so stable together that taking the time to restart things is no big deal. Now having said that, if someone says "for me, restarting aolserver takes 2 days from when I tell it to bounce", then I think there's a problem. I don't expect to ever hear that :) > If you're doing so rarely, then having to bounce aolserver, too, is no big > deal. That's what I thought too; let's see if there's a real use case for this... -Jim ------------------------------------------------------------------------------ Virtualization & Cloud Management Using Capacity Planning Cloud computing makes use of virtualization - but cloud computing also focuses on allowing computing to be delivered as a service. http://www.accelacomm.com/jaw/sfnl/114/51521223/ _______________________________________________ aolserver-talk mailing list aolserver-talk@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/aolserver-talk
