About the waiting for seed State

In TimesTen Scaleout, there could be a situation where an entire replica set fails simulataneously. If all elements in a replica set fail at the same time, it may not be possible to reload any of the elements without unloading the entire database and reloading it. In addition, a reload may cause TimesTen Scaleout to discard transactions against other replica sets that were committed prior to the unload operation.

For more information about replica set failure in TimesTen Scaleout, see Recovering from a Down Replica Set in the Oracle TimesTen In-Memory Database Scaleout User's Guide.

If this case occurs in your Kubernetes environment, the default behavior is for the TimesTen Operator to detect this situation and forcibly unload and reload the database when the situation occurs. This operation is triggered when the Operator notices that all elements in a replica set are in the waiting for seed state.

You have the option of controlling the behavior of the Operator when all elements in a replica set are in the waiting for seed state. You do this by using a TimesTenScaleout object's .spec.ttspec.replicaSetRecovery datum, and setting a particular value for the datum. Accepted values are the following:

• Restart: The Operator forcibly unloads and reloads the database when a total replica set failure occurs. This is the default.

• Manual: The Operator changes the state of the TimesTenScaleout object to ManualInterventionRequired when a total replica set failure occurs. The Operator takes no further action to repair the grid. You must repair it. See About the ManualInterventionRequired State and Set reexamine Datum for details.

For more information on the .spec.ttspec.replicaSetRecovery datum for a TimesTenScaleout object, see TimesTenScaleoutSpecSpec.