AutoSys Workload Automation

Hello,

In Autosys we have a HA Dual Server mode setup.  The primary scheduler and the tiebreaker reside in one data center while the secondary scheduler resides in another.  During our last BCP / DR test we shut down the data center with the primary and tiebreaker.  At this point the secondary scheduler would not work and I presume it is due to the fact that the tiebreaker could not communicate with it.  Does anyone know how to make this setup work?

thanks,

Jim

Hi Jim

Here’s the information from the documentation, and below it is our interpretation. For some unknown reason, I cannot find where it is recommended to put the tiebreaker in the same data center as the shadow, but from what we can discern that is where the tiebreaker should be.

If the connection to one of the event servers (databases) is lost, CA Workload Automation AE does the following The shadow scheduler tries to reconnect to the event server for the configured number of times. If the shadow scheduler cannot reconnect, it rolls over and notifies the tie-breaker scheduler and the application server. The shadow scheduler then checks for status updates from the primary and tie-breaker schedulers. If the primary and tie-breaker schedulers have updated the event server, the shadow scheduler continues to run. If neither the primary scheduler nor the tiebreaker scheduler has updated the event server in two consecutive poll intervals, the shadow scheduler shuts down. If only the primary scheduler has not updated the event server in two consecutive poll intervals, the shadow scheduler fails over and starts processing events.

Our interpretation:

The shadow scheduler tries to reconnect to the event server for the configured number of times.  If the shadow scheduler cannot reconnect, it rolls over to the secondary database and notifies the tie-breaker scheduler and the application server. The shadow scheduler then checks the database for status updates from the primary and tie-breaker schedulers. If the primary and tie-breaker schedulers have updated the secondary database, the shadow scheduler continues to run. If neither the primary scheduler nor the tiebreaker scheduler has updated the secondary database in two consecutive poll intervals, the shadow scheduler shuts down. If only the primary scheduler has not updated the secondary database in two consecutive poll intervals, the shadow scheduler fails over and starts processing events.

Thanks,

Lisa

An excellent dissertation - thanks Lisa!

As a corollary, please consider the version of Autosys that this configuration applies to.

The later versions maintain a separation of the polling arrangement for the EP & ES.

You can architect a layout that uses the native database HA [example Oracle RAC and/or Dataguard] and still use the EP in Autosys HA mode.

Best,

Chris  <CJ>