Re: production server down - Mailing list pgsql-hackers

From Joe Conway
Subject Re: production server down
Date
Msg-id 41C4F6AA.9030002@joeconway.com
Whole thread Raw
In response to Re: production server down  (Andrew Dunstan <andrew@dunslane.net>)
List pgsql-hackers
Andrew Dunstan wrote:
> In the absence of that, in your case, certainly the root-owned 
> placeholder is a good idea - it seems nicer than disabling on-boot 
> startup altogether if you can avoid that.

I'm pretty well convinced at this point that a start on boot init script 
is inappropriate when working with NFS attached storage. We really do 
need to be sure the mount is OK before starting the database. And we 
have the watchdogs in place, and people on call at all times, 
specifically to deal with "service down" events.

And while we could always modify the init script to rip out the 
"helpful" initdb part, I think the under-the-mount safety file is a 
must, unless of course someone thinks of an alternate scenario where it 
will do more harm than good ;-).

Joe


pgsql-hackers by date:

Previous
From: Andrew Dunstan
Date:
Subject: Re: buildfarm improvements
Next
From: Bruce Momjian
Date:
Subject: Re: Shared row locking