[asterisk-users] DID failover

[Jay R. Ashworth]

On Sun, Oct 15, 2006 at 02:06:51PM -0500, Rich Adamson wrote:
> Todd- Asterisk wrote:
> >I'm setting up an asterisk server where an administrator will not always 
> >be available in case of problems.  While I expect problems to be rare, I 
> >need to be prepared.  We're thinking of VoIP DID's and SIP phones so 
> >it's an all TCP/IP network.   We could get a second server to substitute 
> >- What is involved in 'transferring' or 're-registering' the DID 
> >incoming lines to a second server in case the primary is down? If there 
> >a better fall-over method?  I'm looking for the easiest way for the 
> >un-educated sys-admin-apprentice to handle it.   The system doesn't 
> >exist yet so any suggestions are appreciated.   I recognize I'll need to 
> >modify the SIP phones- I'll figure that out later.
> > thanks in advance
> 
> One of the simplest ways to accomplish this is to use an APC power strip 
> with SNMP control. (Each of the power outlets on the power strip can be 
> turned on / off remotely via an snmp command.
> 
> With this rough approach, stop the 'broken' asterisk server and start 
> the backup server (via the power strip control), and wait for the system 
> to come up.
> 
> If both asterisk systems are configured absolutely the same (eg, same 
> *.conf entries, ip addresses), then when the system comes up, it will 
> 'register' with your sip or iax provider.
> 
> The sip phones will likely take a little bit longer to come up due to 
> arp cache timout values within the sip phones. I've not tested any of 
> the sip phones to see what the default timeout values have to be, but it 
> will vary by manufacturer. (Microsoft PC stuff is generally around two 
> minutes.) As soon as that cache value timeouts out, the sip phone will 
> register (with the new server) and should be totally functional.
[ ... ]
> If you read over some of the archives, there are other ways that involve 
> redundant servers, heartbeats, load sharing, reserving a valid extension 
> number that would kick of scripts (etc) to swap boxes. Each have their 
> advantages, disadvantages, and costs.

...and what Rick describes is the little baby version of failover.

Setting up the High-availaibility heartbeat daemons is *reportedly* not
that bad, and will save you the ARP delay by the simple expedient of
having the backup machine take over the primary's MAC address as well
as IP address.

I myself will be needing shortly to look into whether it's possible to
carry other active-session state from one machine to the other (I'm
hoping for transaction-mirrored PgSQL databases on the two machines)
for a project I'm considering.

Cheers,
-- jra
-- 
Jay R. Ashworth                                                jra at baylink.com
Designer                          Baylink                             RFC 2100
Ashworth & Associates        The Things I Think                        '87 e24
St Petersburg FL USA      http://baylink.pitas.com             +1 727 647 1274

	"That's women for you; you divorce them, and 10 years later,
	  they stop having sex with you."  -- Jennifer Crusie; _Fast_Women_