Restart During Run Task Operations
Connect:Direct® for Microsoft Windows provides checkpoint/restart capabilities with the run task Process statement. If a run task operation is executing on the SNODE and a session failure occurs, the PNODE recognizes the session failure and puts the Process in the Timer queue for retry. The SNODE, however, is still running the run task operation and is not notified of the session failure until the operation finishes. The checkpoint/restart feature for run task ensures that when the Process restarts on the PNODE, the run task operation does not execute a second time on the SNODE.
Restart Process Operations
If a run task Process step restarts, the node where the operation executes attempts to find the checkpoint information in the TCQ header. If the run task step is still executing, the Process that is running for the restart of the step waits for the run task operation to finish the first task and proceed to the next step of the Process, if there is one.
When the first Process finishes, it determines that the session under which it was running has been lost and terminates without logging statistics records that indicate a session failure. The second Process records how the run task step that was still executing ended and proceeds to the next step in the Process.
Determine Reexecution of the Run Task Step
If Connect:Direct determines at restart that the run task operation ended because it finished before the PNODE restarted the Process, then the run task step does not execute again. However, if the run task operation did not finish and is not currently running, then the value of the runtask.restart initialization parameter determines whether to restart the Process.
If runtask.restart=yes, Connect:Direct executes the program again. If runtask.restart=no, the Process skips the run task step.