[Asterisk-Users] Asterisk redundancy

[Aaron Daniel]

> I'd like to start a discussion about Asterisk redundancy. I know this has
> been covered in the past, but would like to get an idea of what people are
> doing for a production system that must be up "all" the time.
I'm going to pipe in on this one.  Asterisk redundancy is a huge 
discussion on this list at LEAST once a month, so you would think we would 
have all the details all hammered out and on the wiki and say "Hey, we had 
that discussion last week... what we came up with is on the wiki now."

> HA Linux between the two asterisk boxes. But I am not sure how the Asterisk
> DB would handle a fail over. What happens to the SIP registrations? Can the
> Asterisk DB be offloaded to MYSQL for example? The local DB is importatn
> because this is a call center with agents logged in to multiple queues.
> Config could either be realtime or duplicated manually. What about recorded
> message, has anyone had any problems with an NFS volume providing recorded
> messages such as periodic messages in queues? This solution would require a
> manual swap of the E1 cable inthe event of failure.
Not sure how the agents would work out, you may need a server that handles 
the queueing setup.  We've got 5 servers here (one voicemail, two call 
servers, and two gateways).  The way it's configured, any DB information 
is replicated through a series of dialplan magic and scripts, so I'm not 
sure what it would require to replicate agent information.  I will tell 
you straight up that NFS mounted volumes will cause asterisk to croak if 
it needs access to something that's not mounted.  The first time the NFS 
share disappears for a moment, you're going to be restarting services and 
losing time on the asterisk machines that need the mounts.  It would be 
better to drop the files on all the systems so you don't have to worry 
about that.

>
> Is anyone using a PRI to Ethernet bridge, or any other kind of E1 GW that
> would allow failover to an alternate Asterisk box without manually switching
> the cable? This one is a litte
> expensive(http://www.mapleleaf-technologies.com/webstore/ethernetbridges.php
> ), but seems like it would do the trick. But I would have to run TDMoE
> between the Asterisk boxes and the bridge. Not a big deal probably, but I
> have no experience with TDMoE.
My experience of TDMoE is limited, all of our servers talk sip when a call 
is moved from one box to another.  We have double T1 lines per gateway 
server, so if one goes down, we move to half capacity.  Any calls in 
progress on the one gateway die, but everything else automatically moves 
to using only the other gateway.

>
> I would appreciate any comments regarding redundancy, and how people are
> solving these problems.
>
> Regards to all,
> Joe

-- 
Aaron Daniel
Computer Systems Technician
Sam Houston State University
amdtech at shsu.edu
(936) 294-4198