Connect:Direct Workload Recovery

Edit online

For v6.0 and before when a Connect:Direct/Plex Manager terminates abnormally, or shuts down, the Servers also terminate and are no longer available to execute processes.

With release 6.1, in a situation when Connect:Direct/Plex Manager terminates abnormally, or shuts down:

Connect:Direct/Servers are still available to execute any running processes
Processes continue to run uninterrupted
Statistics could be queried on other Node
Process execution status could be queried on other Node
You will not able to login to Connect:Direct/Plex Environment. Login into ESF mode is however possible.

You will still be able to submit processes and execute them from other Nodes to Connect:Direct/Servers (as SNODE) directly.

Connect:Direct/Servers can be added as Alternate Communication Address (ALT.COMM) in Connect:Direct/Manager entry of PNODE’s Netmap. Example Netmap entry with Connect:Direct/Plex supporting two server. Here, one is added as ALT.ADDR and other as ALT.NODEDEF in Connect:Direct/Plex Manager Netmap entry in PNODE.

$$DELETE
    ADJACENT.NODE=(( CDZSRV1))       
$$INSERT
    ADJACENT.NODE=(( CDZSRV1,30003,10.1.1.21,TCP)  - /* Server 1 node entry */
    PARSESS=(2 1)  -
    SESS.SNODE.MAX=1              -
    TCPAPI=(30004,10.1.1.21) -
    APPLIDS=(M1CDI7P6 M1CDI7P7 M1CDI7P8) -
    )
$$DELETE
    ADJACENT.NODE=(( CDZMGR))
$$INSERT
    ADJACENT.NODE=(( CDZMGR,30001,10.1.1.21,TCP)  - /* Manager node entry */
    PARSESS=(2 1)  -
    SESS.SNODE.MAX=1              -
    TCPAPI=(30002,10.1.1.21) -
    APPLIDS=(M1CDI7P6 M1CDI7P7 M1CDI7P8) -
    ALT.COMM=(                -
      (ALT.NODEDEF=CDZSRV1) -                            /* Server 1 as NodeDef */
      (ALT.ADDR=10.1.1.21 ALT.PORT=30005 ALT.TYPE=TCP) - /* Server 2 as Address */
     ))

When a Connect:Direct/Plex Manager restarts:

Manager syncs-up with all supporting Servers. Statistics Data (STAT), Process Data (TCQ) and Checkpoint Data (CKPT) will be re-synced.
Users can log into Connect:Direct/Plex Manager and query the in-execution or completed processes status that were still running when Manager had shut down
- All active process executions will be continued and can be queried on Manager
- Statistics for all the completed processes can be queried on Manager
Processes that started on Servers (as SNODE) after the Manager terminates or shuts down abnormally, will continue to be executed and their status can be queried on Manager.
Statistics for processes that started on Servers (as SNODE) after Manager terminates or shuts down abnormally, and completed before the Manager restarted, can be queried on Manager.
Note: It is recommended to use Extended Recovery setup with Connect:Direct/Plex Manager so that Standby Manager takes over immediately as soon as the Manager shuts down. This is required to avoid load on Server’s physical memory as syncing data might accumulate over a period.
You can still submit processes and execute them from other Nodes to Connect:Direct/Servers (as SNODE) directly.

Connect:Direct/Plex Manager should be restarted with TCQ initialization parameter set to WARM else Manager/Servers could abend as TCQ will be synced with Servers during Manager restart.

When a Connect:Direct/Plex Manager/Server restarts, its associated servers/manager will issue SITA807I/SITA805I WTO messages to the job log. Also, when a Connect:Direct/Plex Manager/Server shuts down, its associated servers/manager will issue SITA808I/SITA806I WTO messages to the job log. For more information, see Connect:Direct/Plex Status on joining and leaving XCF group.