| 摘要 |
A fault tolerant availability meter includes agents for stand-alone computers and each node of a cluster. The agents monitor availability with timestamps and report uptime and downtime events to a server. Additionally, agents on nodes of a cluster monitor cluster, node and package availability and cluster configuration changes and report these event to the server. Events are stored locally on the stand-alone computers and nodes, and additionally, on the server. Events are tracked with a sequence numbers. If the server receives an out-of-sequence event, an agent-server recovery procedure is initiated to restore the missing events from either the agents or the server. The server may generate availability reports for all monitored entities, including one or more stand-alone computers and one or more clusters of computers. Availability is distinguished by planned and unplanned downtime. Furthermore, unavailable and unreachable systems are identified.
|
