Re: XMin Hot Standby Feedback patch

On Tue, Feb 15, 2011 at 11:42 AM, Simon Riggs <simon@2ndquadrant.com> wrote:
> Patch attached, no docs yet, but the patch is clear.
>
> I'm looking to commit this in next 24 hours barring objections and/or
> test failures.

Looks pretty good to me, though I haven't tested it.  I like some of
the safety valves you put in there, but I don't understand this part:

+                       /*
+                        * Feedback from standby should move us
forwards, but not too far.
+                        * Avoid grabbing XidGenLock here in case of
waits, so use
+                        * a heuristic instead.
+                        */

What's XIDGenLock got to do with it?

On another disk, I think that those warning messages are a bad idea.
That could fill up someone's disk really quickly.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company