Enhancing Scalability with Master Scheduler

This topic describes the behavior of the Process Scheduler sever in various fail over situations to provide you with some reasonable expectations of scalability in comparable situations.


System Behavior

No Master Scheduler configured or running.

  • No load balancing will occur.

  • A process will be run by the designated Process Scheduler server only.

  • A process will remain queued until the designed server is up or it is capable of taking requests.

  • A process assigned to an overloaded server will not be transferred to other idle or free servers.

One or more Master Scheduler servers running.

  • Load balancing will be available as per configuration.

  • If more than one Master Scheduler is configured, the system is guaranteed to have one Master Scheduler available.

Two or more Master Scheduler servers are configured.

If one Master Scheduler is down, the other will become master scheduler automatically

Master Scheduler available and one of the servers is overloaded.

Work load on overloaded server will be distributed to other servers as per configuration.

Master Scheduler is available, and one Process Scheduler server is brought down, while processing a request.

  • There will not be any impact to a running process as it will not be stopped abruptly.

  • In case of Application Engine, and AESRV is enabled, Process Scheduler shuts down gracefully, so, there is no impact on the running process.

  • Reports will not be posted on completion of the process if the Process Scheduler server is brought down while running a process. However, reports will be posted if the same Process Scheduler is booted.

Note: The items above assume that the Process Scheduler severs is shut down using PSADMIN, using the normal shut down option, not the forced shut down option. If the forced shut down option is selected, the running process (running at the time of shutting down) will be killed.

Process Scheduler is brought down when it has a number of processes queued for processing.

Queued processes will be transferred to other available servers depending on configuration.

Note: The above items assume that web server is always up. If it is down, report distribution can be disrupted.

Note: Abrupt/forcible killing of servers using utilities like Task Manager of forced shut down using PSADMIN may lead to irrecoverable change of state of a running processes.

Note: All configured Master Scheduler servers should not shut down at the same time. The above statements assume that, at any given time, at least one Master Scheduler server is up and running.