GRACEFUL ROLLOVER

Staring with Release 8.1.400.??, URS simplifies the process for moving call processing from one URS instance to another URS. Once a working/active primary/backup pairs of URSes exist, URS automatically performs all necessary steps for a graceful rollover using those relationships.

How It Works

When instructed to start graceful rollover, Management Layer still sees the primary/backup pair relationship between the two URSs, but internally those URSs turn into a cluster of two independent URSes, which stop to execute the graceful rollover commands. Graceful shutdown is not applicable for a single URS instance with no backup.

Prerequisites

The prerequisites for a graceful rollover are as follows:

• A primary/backup URSes must be configured and work.
• As the graceful rollover process temporary turns a primary/backup pair of URSes into a cluster of two independent URSs, Genesys recommends configuring agent reservation before using graceful rollover.
• If Orchestration Server backed up calls are present, verify that the current backup URS "sees" all such calls. This can be achieved with a web request to the primary URS: urs/console?backup. The response will be “name(host:port,socket);0000000000000000”, The presence of a not null connection id after the semicolon indicates that it is not safe to start the rollover process as some ORS backed up calls can be lost..

Tip

.

.

Graceful Shutdown Command

Graceful shutdown is started either with web request to current primary URS urs/console?rollover or with console command rollover or through Local Control Agent command Suspend (Graceful Stop) .

When graceful shutdown is started with command rollover

      - if there is no backup configured then URS will reply with no backup and will not start graceful shutdown procedure
      - if URS has no connection with its backup then URS will reply with no connection and will not start graceful shutdown procedure
     - URS will reply with name of its backup server and starts graceful shutdown procedure. 

When graceful shutdown is started through LCA URS will just ignore request if there is no proper backup. Effectively it will result in not graceful shutdown as SCS in such case will request application to quit.

When graceful shutdown command is received:

 - current primary URS stops processing of new interactions (from TServers or Web clients). 
      It stops processing of any multimedia calls (returns them back into Ixn queues) and ORS backed up calls.
      It imitates switching into backup mode of operation forcing ORS (if any) to switch to current backup node.
- current backup URS delete all interactions (as they all will be processed by current primary node) excepts ORS backed up ones. 
     It switches into primary mode of operation and start processing of new interactions as they arrive.
- when original primary finish processing of all existing calls it automatically exits.

Detailed Design