Engine Startup and Shutdown Sequence

Engine Startup and Shutdown Sequence
This section outlines the main actions that occur during engine startup and shutdown (in normal circumstances). In any particular project only some of the actions may be required. For example, a project may have no startup rule functions.
TIBCO BusinessEvents Express  Content relating to Cache OM and backing store does not apply to TIBCO BusinessEvents Express edition.
Except where noted, this section assumes cache OM and inference agent startup and shutdown. It provides the main milestones only.
Cluster Startup and Shutdown
There are only two main points to keep in mind for orderly system startup and shutdown:
Start storage-enabled agents (cache agents) first   When Cache OM is used, you must start a node that has storage enabled first. In production systems that would be a dedicated cache agent engine. (In test deployments this could be another type of agent node with cache storage enabled.)
Stop other engines before storage-enabled agents (cache agents)  In unusual situations where all cache agents are stopped but engines running other types of agents are running, you must restart all engines.
Startup Sequence
During engine startup the following actions occur:
1.
If persistence OM is used, the location of the Berkeley DB software it is using, and information about what was recovered from the database.
2.
Cache OM with backing store only: Recovery stage. When the minimum number of cache agents is started (as defined by the Cache Agent Quorum CDD setting), the cluster enters the recovery state. Various caches are preloaded from the backing store, according to preload settings. When Recovery state ends, the cluster enters Ready state.
3.
All inference agents build their Rete networks by evaluating conditions against all Cache Plus Memory objects (if any).
(Cache OM only) Inactive (Standby) Nodes  If all agents in an engine are inactive, then this ends the startup sequence for that engine.
4.
5.
6.
7.
The first RTC cycle occurs and all rule actions that are eligible to execute now execute. (Scorecards and startup rule functions can cause rules to be eligible to execute. Depending on the state of entities recovered from the backing store, the RTC will take more or less time.) See TIBCO BusinessEvents Architect’s Guide for more details about RTC cycles.
8.
9.
Rule-based time events (recovered or scheduled in a startup action) are asserted after the specified delay. The delay begins when the rule or rule function action executes, so at startup, it is possible for time events to have passed their start time, and they are asserted immediately.
10.
Shutdown Sequence
During engine shutdown the following main actions occur:
1.
2.
3.
4.