* Simon Riggs (simon(at)2ndQuadrant(dot)com) wrote:
> I get the feeling that part of the inspiration for this is that Hot
> Standby must maintain this file. If not, I'm curious as to the reasons
> for doing this. No objections however, just curiosity.
The impetus for these changes was the performance complaint that adding
new users is extremely expensive once the files get to be large.
Solving that kind of a problem typically involves moving away from flat
files and to a database. Thankfully, we happen to have a database
already built-in. ;)
> Specifically, should I remove the parts of the HS patch that refresh
> those files?
I'd probably wait till it's committed and then rip it out as part of
re-baseing against CVS. That's just my 2c on it and I've no clue how
invasive the changes to HS are to do that. If it's easy to separate the
changes and remove/add them as needed, then perhaps it doesn't matter if
you do that now or wait.
In response to
pgsql-hackers by date
|Next:||From: Tom Lane||Date: 2009-08-29 15:06:44|
|Subject: Re: WIP: remove use of flat auth file for client authentication |
|Previous:||From: Simon Riggs||Date: 2009-08-29 11:02:02|
|Subject: Re: LWLock Queue Jumping|