On Mon, 2008-11-03 at 17:37 -0500, Greg Stark wrote:
> Secondly the locking seems to be a bit overoptimistic. I'm pretty sure
> you have to take an exclusive lock on an index page any time you make
> any data modifications in index pages -- even if you're just setting a
> bit and not moving any data around. If two processes set two bits in
> the same word one can get lost in the race condition.
I looked at that aspect of the patch specifically a few weeks back while
checking for possible issues with Hot Standby. IIRC the patch is fairly
careful with locking and uses Exclusive locks extensively throughout. I
looked at both the theory and the implementation. Unless Gianni changed
something in that regard recently, I don't understand that comment at
all. Probably need to provide specific examples of your concerns.
> There are a lot of comments in the code which imply that vacuuming is
> not implemented but in fact from what I can see it is -- sort of. It
> does rewrite the bitmap in bmbulkdelete but it doesn't have to rebuild
> the index from scratch. Are the comments out of date or am i
> misunderstanding them or the code? How complete is the vacuum
> implementation?
As I understood it, complete. I think the objective was minimal change
away from Gavin's original. But it sounds like you found some out of
date comments.
Extensive docs have been added as a README, mainly because it was pretty
hard to understand without them.
-- Simon Riggs www.2ndQuadrant.comPostgreSQL Training, Services and Support