vicidial.org

Vicibox 6.0.4 from .iso | Vicidial 2.12-538a Build 160122-1401 | Asterisk 1.8.32.3-vici | CLuster setup: 1 web 1 DB 5 telephony | No Digium/Sangoma Hardware | No Extra Software After Installation


what would be the best practices for using untangles WAN Failover with a clustered vicidial setup that has 5 dialing nodes that are NAT-ed? When one WAN fails and the other kicks in we lose all dialing capability since the nodes are set to the ips of the failed WAN. Any ideas?

Don't use WANs that fail. The SIP Protocol is not designed to survive such a breach.

Failing that, you could attempt to configure a VPN connection through your funky configuration that would ultimately view the traffic pattern differently and allow a sound pause and continue. But the power needed to keep the sound quality over VPN likely requires dedicated hardware.

But more to the point: This is a PBX environment. It's like asking how to run a taxi service with cabs that all have bicycle tires. You're not likely to enjoy the experience if your internet is not consistent.

Perhaps you should look at moving your Vicidial servers to a colocation facility with reliable internet. Then VPN the agents to the colocation facility if necessary. First off, you'll be dramatically reducing your internet usage locally, which may allow it to stabilize. Secondly, with that reduced bandwidth, VPN may become viable due to the lower usage.

I wonder if the Untangle is not the actual problem. If you have not tried this with simpler physical routers instead of untangle, you're likely overlooking what could turn out to be a viable solution. We've found that "special" routers are often the problem and simply taking them out and putting in a router from Walmart has solved many client's problems for Call Center operations in the past. Especially with Untangle and pfSense. Just sayin'

we solved this by using one isp outside of untangle and the other isp through untangle but on 1:1 NAT. This way if the connection fails we can switch to the other with limited config change and can resume dialing in under 5 mins. Thank you for you help William!

You could just build a script that checks the network connections every minute and reconfigures the carriers if the primary is down. Not very different from a DNS check or any one of a myriad other system checkers.

Thank you william. We ended up building 2 identical systems (1 in each location) in which 1 is on one isp and the other is on the backup isp. We set all the web address for agent UIs and softphones using DNS which translates to each node and is controlled by untangle. It works really well with network outage and/or hardware failure(experienced both already)and all it involved was switching where the DNS pointed and having the agents log out and back in and it was back to dialing.

TwistedFister wrote:Thank you william. We ended up building 2 identical systems (1 in each location) in which 1 is on one isp and the other is on the backup isp. We set all the web address for agent UIs and softphones using DNS which translates to each node and is controlled by untangle. It works really well with network outage and/or hardware failure(experienced both already)and all it involved was switching where the DNS pointed and having the agents log out and back in and it was back to dialing.

Excellent postback!

Do you see room for improvements anywhere? Are you using local DNS so you can update it immediately? Also, for future expansion, be sure the web DNS and phone DNS are not the same. When you separate the web/dialer functions this will come in handy. Also consider max 10 agents per DNS group so you can easily move agents from server to server in groups of 10 during later expansion. Enterprise Concepts.