On Tue, Apr 22, 2014 at 11:11:29PM +0530, Kashyap Chamarthy wrote:
On Tue, Apr 22, 2014 at 08:34:58AM -0500, Ryan O'Hara wrote:
> On Tue, Apr 22, 2014 at 11:56:49AM +0530, Kashyap Chamarthy wrote:
[. . .]
> > Please feel free to correct/comment if you see something off the mark.
>
> Need to set client/server timeouts for the qpid proxy. This should be
> no less that 60, but depends on your qpid heartbeat. 120s is probably
> a good starting point.
/me hasn't configured haproxy before, just wondering, I assume this is
section in haproxy.conf you're referring to where I set the qpid
timeout:
[. . .]
frontend vip-qpid
bind 192.168.16.201:5672
default_backend qpid-vms
[. . .]
Yes. That is half of it, at least. The default client/server timeout
is 10s, set in the "defaults" section. You can override that per
proxy. You want the client timeout in the frontend and the server
timeout in the backend. Actually, the mysql proxy in your wiki page
does this, setting the timeouts to 90s.
> I think we need to collapse some of the wiki pages. There are
now 2-3
> docs describing how to use haproxy with OpenStack.
Can you please post them here if you have them handy?
This is the one I wrote a while back. It is desperately in need of an
update.
http://openstack.redhat.com/Load_Balance_OpenStack_API
> > > I'll try to make a high-level document which will
include
> > > references to all the wiki pages.
> >
> >
http://openstack.redhat.com/Setting-up-High-Availability
>
> OK. There is also an old(er) doc about how to do this which is
> probably obsolete.
Yeah, if you find them, it'd be useful to mark them as obsolete, just to
avoid confusion.
This is what I was thinking of:
http://openstack.redhat.com/RDO_HighlyAvailable_and_LoadBalanced_Control_...
Thanks for the comments, Ryan.
Sure thing.
Ryan