Other SIP Server HA Enhancements
SIP Server supports several additional capabilities related to high-availability deployments.
- Single Host HA Deployment
- Synchronization of Contact Between SIP Server HA Pair
- SIP Traffic Monitoring
- Monitoring Critical Conditions
Single Host HA Deployment
Starting with version 8.0, SIP Server supports deploying both primary and backup SIP Server applications, as well as the Stream Manager or Media Server application, on the same physical host. Benefits of using the single host HA configuration include the following:
- Efficient use of the hardware equipment.
- Less complex configuration: Virtual IP address control scripts, Alarm Reactions, and Alarm Conditions are not required.
However, this type of HA configuration is supported only for small-size deployments--100 seats or less.
Synchronization of Contact Between SIP Server HA Pair
SIP Server 8.x synchronizes the SIP registration Contact header for a particular device across both primary and backup instances of SIP Server. The primary SIP Server sends the contact information to the backup SIP Server using the HA link, as well as through the Configuration Server.
SIP Traffic Monitoring
SIP Server 8.x supports SIP traffic monitoring for enhanced reliability. When configured, SIP Server monitors incoming SIP traffic and can initiate a switchover after a configurable length of time during which no SIP messages are received.
In deployments where two NICs are used, one NIC is dedicated to SIP communication, while the second NIC is used for other kinds of communication with various components. Solution Control Server (SCS) manages and monitors the SIP Server application through the second NIC.
The SIP traffic monitoring feature allows the primary SIP Server to monitor the network connectivity through the NIC that is responsible for SIP communication, to recognize connectivity issues that impact the SIP service, and to initiate reactions that result in recovery of the service.
An Application-level configuration option, sip-pass-check, must be configured to enable this functionality. In addition, at least one service device must be configured for Active Out-Of-Service Detection by using oos-check and oos-force configuration options. See the Framework 8.1 SIP Server Deployment Guide for information about the Active Out-Of-Service Detection feature description.
When it is set to true, the sip-pass-check option enables tracking of SIP messages that reach the primary SIP Server, including responses from SIP devices (DNs) that are monitored by SIP Server by using the oos-check and oos-force options.
The primary SIP Server summarizes results of the checks on DNs for out-of-service status and monitors the time that has passed since the last received SIP message. If the primary SIP Server does not receive SIP messages for a certain period of time, SIP Server reports the SERVICE_UNAVAILABLE status to LCA/SCS. The period of time is chosen as the maximum of sums (among the sums of the oos-check and oos-force option values, configured for service DNs). When SIP Server reports the SERVICE_UNAVAILABLE status to LCA/SCS, SCS switches the primary SIP Server to the backup mode and this SIP Server reports the SERVICE_RUNNING status to LCA/SCS. The former backup SIP Server becomes the primary server and starts to monitor SIP traffic.
If both the primary and backup servers receive no SIP traffic, a switchover would occur each time that the effective out-of-service timeout expires. To prevent frequent switchovers in this case, SIP Server detects the "double switchover" condition and doubles the effective out-of-service timeout each time that the double switchover happens"up to four times greater than the initially calculated timeout, or until one of the two servers detects SIP traffic. As soon as SIP traffic is detected, the server that detected the traffic remains the primary SIP Server and continues normal operation.
Monitoring Critical Conditions
You can use Genesys Administrator to check the current running status of SIP Server. Starting in release 8.1.0, SIP Server displays its state as Running in Genesys Administrator in cases where it is unable to open a listening port, and it is configured as one instance in a High Availability (HA) pair. Prior to release 8.1.0, (release 8.0.4 and earlier), in this same scenario SIP Server displayed its status as UNAVAILABLE.
To monitor problems with binding a listener (SIP Server is running but unable to open a listening port), Genesys recommends that, for each SIP Server instance, you configure an Alarm Condition for the log event 00-04200. For more information, consult the Solution Control Interface (SCI) help topic, "Using Log Events for Alarm Detection".
To ensure that administrators do not miss the alarm, Genesys recommends that you configure automatic clearing of the activated alarm in accordance with business processes and the schedule of the customer administrator.
The recommended configuration of an Alarm Condition for 00-04200 enables monitoring of a wide range of events that are critical for both SIP Server functionality and for service availability. This includes problems that might occur when binding a listener, unexpected terminations, or unauthorized terminations of the SIP Server process.
In Genesys Administrator, alarms that are detected and activated can be observed through a dedicated view, providing a central location for observing all alarms that occurred in the entire environment.
If required, an alarm reaction can be configured to notify administrators automatically when a critical condition occurs.
After the administrator investigates and resolves the problem, they must manually clear the alarm condition.
If the problem occurred due to a temporary outage (for example, a network switch reboot), SIP Server remains in the Running state, ensuring availability of the HA pair once the network switch is recovered; in release 8.0.4, SIP Serer required a manual restart to return to the Running state.
In release 8.0.4, if both SIP Server instances encountered a problem when binding a listener, both instances in the HA pair remained in UNAVAILABLE status, requiring a manual operation to resume the service. In release 8.1.0, SIP Server instead switches the primary role between the two HA instances and resumes the service as soon as one of the instances is able to open a listening port.