Synchronization of calls in progress with new Backup SIP Server

Iteration 110. FDS: https://intranet.genesys.com/display/RDSIPS/%5BFDS%5D+Synchronization+of+calls+in+progress+to+new+Backup+SIP+Server

Impacted pages:

• https://docs.genesys.com/Documentation/SIPS/8.1.1draft/HADeployment/RedundancyTypes#Hot-Standby_Redundancy_Type
• https://docs.genesys.com/Documentation/SIPS/8.1.1draft/HADeployment/ConfOptionsHA
• https://docs.genesys.com/Documentation/SIPS/8.1.1draft/HADeployment/NewInThisRelease

When a backup SIP Server is restarted or HA link connection between primary and backup was lost and then re-established, some calls would only exist on the primary SIP Server. Previously, SIP Server would not synchronize missing calls. Starting with version 8.1.102.xx, primary and backup SIP Servers, after establishing HA link connection, will exchange information about calls and trigger synchronization of missing calls to the backup SIP Server.

The synchronization of new calls will happen as soon as the HA connection is established. Immediate switchover or failover after HA Link connected should be considered as double failure.

Readiness for Switchover or Upgrade

SIP Server reports readiness to switchover and readiness to upgrade over its http interface. This will allow support of quick automated SIP Server HA pair upgrade with following steps:

1. check Backup SIP Server for readiness to upgrade.
2. upgrade Backup SIP Server
3. check Primary and Backup SIP Servers for readiness to switchover. If there is no multi site calls, the time to become ready in normal conditions (no network interruptions and total number of calls less then 1000) is expected to be 1 minute after Backup SIP Server started. For multisite configuration Primary becomes ready for switchover when all unsynchronized multisite calls are released or 2 hours after HA link was established.
4. Initiate switchover.
5. check Backup SIP Server for readiness to upgrade.
6. upgrade Backup SIP Server

Criteria to report readiness for switchover:

• Primary SIP Server: All calls are synced to Backup and HA link connection established. There are no un-synced multisite calls.
• Backup SIP Server: All synced calls received and HA link connection established.

Criteria to report readiness for upgrade:

• Primary SIP Server. Never ready. It should be switched to backup mode.
• Backup SIP Server. Always ready

SIP Server reports its readiness to upgrade and switchover over its web interface:

SIP Server application option http-port has to be configured on application level. Web Interface can be accedes using http://<SIPS_HOST>:<HTTP_PORT> Readiness to upgrade and switchover available in "Sip Server" section:

Reliable upgrade procedure of SIP Server HA pair:

1. Check SIP Server in Backup role for readiness to upgrade via web interface
2. Stop SIP Server running as Backup. Install new version and start again.
3. Check Primary and Backup SIP Servers for readiness to switchover via web interface and initiate switchover if readiness is 1.
4. Check SIP Server in Backup role for readiness to upgrade via web interface
5. Stop SIP Server running as Backup. Install new version and start again.

ha-max-calls-sync-at-once

Section Name: TServer, Application level
Default Value: 500
Valid Values: 200-1000
Changes Take Effect: When HA link connection is established

Specifies the maximum number of calls that can be synchronized at once from the former primary SIP Server to the new primary SIP Server (the former backup) after the HA link connection is established, before waiting for 1 second to continue (with synchronization?). Only calls that are missing on the former backup SIP Server are synchronized. This option enables to throttle synchronization of calls in case of a slower network connection between primary and backup SIP Servers.

Feature Limitations

In case of multi-site call, information about ISCC links between parties on different SIP Server HA pairs not synced during HA link connected. As result for such call ISCC functionality, in case of immediate Switchover, could be broken. Complicated multi-site call flows which include OOSP transfers can result in a TEP-only (transit) call. Such calls exist only to propagate T-Events and updates between other sites, maintaining integrity of a multi-site call. Such call doesn't have TSCP/DDP counterparts. Thus, such transit calls will not get replicated to backup server after Backup SIP Server restart or reconnect