[Pulp-list] pulp HA

[Steven Roberts]

On Wed, Feb 06, 2013 at 08:58:51AM -0500, Jay Dobies wrote:
> On 02/06/2013 02:35 AM, Steven Roberts wrote:
> >As I am looking more at pulp, one of the thoughts that has crossed my
> >mind is HA capabilities for deploying pulp.
> >
> >Specifically to the two datacenters we have.
> >
> >I'm thinking active-standy wouldn't be too bad to do now.  use the
> >Mongo replication features I have heard about and do an rsync of
> >the /var/lib/pulp dirs.
> >
> >For more of an active-active not sure if there are any roadmap plans
> >for this.
> 
> Can you give some more information on what you mean by active-active
> here? Are you talking about two different Pulp servers serving the
> same repo at the same URL (not sure if that falls more under the
> idea of load balancing than HA, but at that point, we're just
> talking semantics)?

Assume two datacenters, A and B

A-pulp -> pulp server at datacenter A
B-pulp -> pulp server at datacenter B
A-servers -> several hundred linux servers needing yum repos for
             installs,updates at datacenter A
B-servers -> several hundred linux servers needing yum repos for
             installs,updates at datacenter B


Active/active:
A-Servers use A-pulp normally and can failover to B-pulp if A-pulp goes down.
B-Servers use B-pulp normally and can failover to A-pulp if B-pulp goes down.
A-pulp and B-pulp synchronize their config & data.

active/standy:
A-Servers use A-pulp normally and can failover to B-pulp if A-pulp goes down.
B-Servers use A-pulp normally and can failover to B-pulp if A-pulp goes down.
A-pulp synchronizes its config & data to B-pulp.

I can make a common domain name that resolves correctly no problem.  it
is are there mechanisms to have the pulp-servers stay in synch.

Steve