--- Jan Wieck <JanWieck at Yahoo.com> wrote:

> On 1/31/2006 9:25 PM, Ujwal S. Setlur wrote:
> > Hi,
> > 
> > My application generally connects to the database
> > using a specific user. This user does not have
> view
> > permissions on the slony schema. However, my
> > application needs to look at some slony
> information to
> > make some decisions. I also do not want the
> > application to use the postgres super user
> account.
> > 
> > Is it OK to give the application user "read only"
> > privs on the slony schema? I can't imagine that
> would
> > be a problem, but I thought I would ask.
> 
> I don't see any problem with giving your application
> read permission on 
> certain slony specific tables.
> 
> That said, slony does not change the ownership
> (origin) of sets by 
> itself. Some mechanism you have setup yourself
> explicitly instructs 
> slony to switchover or failover. Can't you
> incorporate telling your 
> application where the current origin is into that
> process? I have done 
> that in my test cases usually with replacing the
> pgpool config file 
> during switchover or failover.

You are right, I can do that. I do have a script that
does the switchover/failover. The issue, though, is
that:

1. The server side code actually consists of 3 daemon
processes that interact with the db, and they all need
to know if a switchover/failover has occured.

2. When the daemons are shutdown and restarted, they
must wake up with the knowledge that a
switchover/failover has already occurred, so that they
know which node to connect to.

I am currently not using pgpool. I am not familiar
with it. Maybe I should explore it?

Thanks,

Ujwal

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com