By Date: <-- -->
By Thread: <-- -->

HA: setup instructions



Martin,

This is a great topic for HA 2.0 (by which I don't mean anything just specific, just a label for whatever comes next). HA 1.0 is proving to be a  challenge so we need to stay narrowly focussed with the execution -- but it's good to think forward to where we are headed from here. Creating a good set of use cases will make sense. Another requirement for the next rev will be to provide an integrated installation procedure rather than manual hacks to the single-server install. We also need to generalize from 2 registrars to N, with N remaining a relatively small number, at least for a while.

-Walter

Scott Lawrence wrote:
On Mon, 2006-01-23 at 10:46 -0500, Martin Steinmann wrote:

  
First draft of setup instructions checked in to sipXregistry/doc on
      
the HA branch, as HaSetup.odt (OpenOffice) and HaSetup.pdf. Please
review. In particular, >please comment on the sections that I have
included below.
    
-Walter
      
Walter

Very interesting read. How about defining some specific use cases for
which install instructions are provided?  Here is what I am thinking
of:

1. Redundant HA setup in the same data center with both servers on the
same subnet. Major objective: Provide redundancy and failover for call
control.
    

It actually doesn't matter one way or the other whether they are on the
same subnet.  They just need to have IP connectivity - you could sync
two servers on opposite sides of the planet if they can get https
connections between them.

In fact, I would expect the normal case to be servers in different
locations because it makes the failures less likely to be common.
Backhoes can be tough on a single site.

  
2. Two location setup. The two servers are on two different subnets
connected over a VPN (i.e. two locations). Each server is primary
responsible for its population of phones and gateways (i.e. the phones
and gateways in that location). If the server in one location goes
down, the server deployed in the other location would take over.

The second example is what we probably would like to do. The two
locations should be independent both having their own PSTN T1 gateway.
It should be possible to make and receive calls even if the IP
connectivity to the other location is broken. Under normal operating
conditions the two systems would look like one big system and phones
could have DID numbers from both locations registered.

Would this make sense?
    
 
It would make sense, but it's beyond the scope of what we're trying to
achieve for this release.   Might be possible with some extra hand
configuration though.

  
_______________________________________________
sipx-dev mailing list
sipx-dev (at) list.sipfoundry.org
List Archive: http://list.sipfoundry.org/archive/sipx-dev