[Pgpool-hackers] [Pgpool-general] pgpool limitations

[Gurjeet Singh]

On Tue, Aug 2, 2011 at 7:46 AM, Tatsuo Ishii <ishii at sraoss.co.jp> wrote:

> > IMHO, it should be possible to remove a slave backend and disconnect
> _only_
> > those clients which have active queries on the slave going down, while
> other
> > slaves are still connected and running their queries. I understand that
> in
> > Master/Slave configuration every incoming connection has outgoing
> connection
> > to a master and a slave, but it'd be great to allow clients to stay on
> while
> > an unrelated slave is being brought down manually.
>
> That would be great and that has been on my personal TODO list for
> while. I just don't have time to deal with it now.
>

Good to know that :)

I have tried to look through the code and see if I could figure out what
needs to be done, but couldn't make much progress.

I think what we need to do is, in failover(), instead of killing all
frontend precesses we should kill only those that have active connections to
the node we are taking down.

I am not able to figure out how to find all frontend connections that are
connected to a given backend node-id.

Regards,
-- 
Gurjeet Singh
EnterpriseDB Corporation
The Enterprise PostgreSQL Company
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pgfoundry.org/pipermail/pgpool-hackers/attachments/20110802/ddd99b35/attachment.html>