Child pages
  • SipXbridge and Interoperating with High Availability ITSP Nodes
Skip to end of metadata
Go to start of metadata

This document describes the issues and workaround of connecting to a high availability ITSP node using the SipXcom SipXbridge pseudo-SBC capabilities. The high-availability ITSP node configuration provided by Ace Innovative will be used to describe the situation - the network configuration is as follows:



Ace Innovative has recently introduced a high availability service and has begun migrating most of their clients to this ITSP node named csp1.acepbx.com. There are actually two target nodes behind csp1.acepbx.com - csp1.acepbx.com (66.114.83.74 ) and csp2.acepbx.com (66.114.83.75). csp1.acepbx.com is provisioned in the address field  (PSTN gateway address in the Devices->Gateways->SIP trunk menu) when the SIP trunk is defined. The way Ace has defined the service is that all traffic goes through the primary csp1 node and the backup csp2 node is used only when csp1 is unavailable. In the above diagram, Ace Innovative has defined a NAPTR record for csp1.acepbx.com called
_sip._udp.srv.csp1.acepbx.com. That NAPTR record is then used to query the SRV records - two are returned. The target node csp1.acepbx.com is the record with the lower priority (10) and Sipxbridge should register to this node. Sipxbridge queries for the NAPTR and SRV records correctly - the issue is that Sipxbridge prior to 17.10 does not process the SRV records correctly, and registers the trunk to csp2.acepbx.com about 50 percent of the time. sipXbridge in 17.10 and later now handles ITSP connections with SRV records.

The good news is that on a busy Sipxcom system, there is no impact when the SIP trunk registers to csp2.acepbx.com. Sipxbridge always sends a keep-alive UDP packet every 20 seconds to csp1.acepbx.com regardless of whether the SIP trunks registers to csp1 or csp2. If there are incoming PSTN calls that arrive within the 10 minute SIP trunk registration window, then the invites will continue to be delivered from the primary csp1 node. The firewall state for csp1 is maintained by the 20-second Sipx keep-alive packet, so the incoming call successfully traverses the firewall to the Sipxcom server. 

If there are no incoming calls within the 10-minute  interval when the SIP trunk is registered to csp2, and the subsequent registration also registers to csp2, then invites for incoming PSTN calls on the SIP trunk will be sent from csp2. This presents a firewall state issue - when the SIP trunk registers to csp2, firewall states are created for csp2 and the 20 second keep-alive maintains the firewall states for csp1. However firewall state information for csp2 ages out in 60 seconds when default firewall parameters are used.

The workaround is to replace the csp1.acepbx.com address in the SIP trunk definition with the actual IP address (64.114.83.74) in the Sipxcom gateway configuration definition - this forces all registrations to csp1 but bypasses the high availability capability.

 

  • No labels