PRIMECLUSTER Shutdown Facility (SF) provides features to forcibly stop nodes where an error occurs in the cluster.
The SF is made up of the following four major components:
Shutdown Daemon monitors the state of cluster nodes, collects the states, and provides the interface to request a shutdown manually or automatically. Also, a processing to solve a cluster partition state is performed.
Shutdown Agents guarantee the shutdown of other nodes. Though Shutdown Agents are attached to SF products, they vary depending on the architecture of cluster nodes for SF installation destination. SF provides the feature to shut down a node whether RMS is operating or not for each product of PRIMECLUSTER service layer.
Monitoring Agent monitors the state of cluster nodes by taking advantage of the characteristic of hardware and detects the node down immediately. When errors occur on other nodes, such as sudden system panic or power off, the error is reported to SF. Also, the feature as Shutdown Agents (SA) is provided to shut down nodes which errors occur.
The sdtool(1M) is the command to provide I/F of Shutdown Daemon.
PRIMECLUSTER Shutdown Facility has the following features:
It is possible to detect the shut down or cluster nodes immediately (monitoring agent).
It is possible to shut down cluster nodes whether RMS is running or not.
It is possible to shut down cluster nodes from any components of PRIMECLUSTER service layer.
The first section explains the initial installation of SF products. The second and subsequent sections explain the configuration setup of SF. The last section explains the changes which need to be added to other products.
