Action descriptors let you track or change an action.
When an asynchronous command seeks to stop one or more instances,
it generates an action descriptor. The action descriptor tracks
the action, the wait option, and target instances in the time_wait
state. You can view information about an action descriptor by querying
the monLogicalCLusterAction table or executing sp_cluster
logical, "show", NULL, action.
An action can be “active” or “complete.” An action is active when at least one target instance remains in the time_wait state. An action is complete when all target instances are no longer in the time_wait state.
Using sp_cluster logical, action, you can manage action descriptors using these options:
cancel – terminates active actions. Instances in the time_wait state due to that action return to the online state. Existing connections remain, even if they were marked for migration or termination.
If the command that resulted in the creation of the action brought instances online, they remain online. For example, if an action results in the cancellation of a failover from s1 to f1, f1 remains online.
modify_wait – changes the wait option (see “Asynchronous commands and logical cluster states”) and time associated with an active action. For example, if an action is created with a 10-minute wait, use modify_wait to change:
The time delay to 20 minutes
The time delay to the actual clock time of 4:30 p.m.
The wait option to nowait
release – removes a completed action from the monLogicalClusterAction table.
Completed actions remain in the monLogicalClusterAction table so you can track their status. However, completed actions consume memory from the workload manager cache. Execute the release command after an action completes to free this memory.
Action information is stored in memory only. Restarting
the full cluster clears all actions from the monLogicalClusterAction table.
