Warm Services
Warm services reduce the time needed to activate a replacement NMS Services instance by loading supported service caches before activation. A warm instance is a non-primary instance: it uses read-only database access and does not serve operational end users. It caches relatively static data; activation still requires the remaining initialization and current operational data to be loaded. Warm services do not provide a second fully active set of NMS Services. Only one set of NMS Services may be active for an NMS instance at a time.
Warm services can be used on a backup application node at the same site or at an alternate disaster recovery site. They must have access to the NMS database for the instance they will take over. At a disaster recovery site, the corresponding physical standby database must be open for read access while replication continues; this is the Active Data Guard configuration described in
Prerequisites. A standby database that is only mounted cannot supply the queries needed to initialize warm services.
Preparing the Standby Instance
1. Make the NMS executables, configuration, and required model files available on the standby application node. A shared filesystem can reduce differences between application nodes at the same site. If files are installed separately, keep the copies compatible with the NMS database and the intended active installation. Warm operation does not replace model-file synchronization or the patching procedure.
2. Configure the standby node's NMS environment, database connection, system.dat, and local communication endpoints. Keep client routing directed to the active instance. If local CORBA services must start before a floating IP address is assigned to the standby node, configure their listen addresses for an address already available on that node. The naming-service settings NMS_NS_LISTEN_ADDRESS and NMS_NS_CONNECT_ADDRESS must agree: for example, listening on 0.0.0.0 permits a local connection through localhost. Select network bindings according to the deployment's network configuration.
3. From the configured standby administrative environment, with no SMService already running in that environment, start warm services:
sms-start -warm
SMService adds -warm to the managed Isis service processes it starts. A service that does not support warm mode stops instead of initializing in warm mode. Processes configured as non-isis are started normally and do not inherit warm behavior. Review custom processes and integrations in system.dat before including them in a standby installation.
4. Use smsReport to inspect the standby instance. Confirm that services supporting warm mode have reached their warm state. Investigate unexpected failures separately from a service intentionally stopped because it does not support warm mode. Confirm that the active instance continues to serve users.
Activation and Return to Standby
For recovery between sites, use the configured nms-ha-switchover or nms-ha-failover workflow described in
Switchover and Failover. These scripts perform the database role transition, attempt the configured model-file synchronization, update the current NMS site, and invoke nms-all-start. The startup script uses
sms-start -f system.dat -hot to promote an existing warm service set and start stopped services. Promotion is coordinated by SMService, which sends the hot message to the running managed service processes.
For recovery between application nodes at the same site, include promotion in the site's cluster recovery procedure, after the previous active instance has been stopped or isolated and the replacement node has the required database and filesystem access. Do not promote individual managed services independently.
After activation, confirm that required services are running, database access is read/write on the primary database, WebLogic is connected to the active services, and operational client and integration checks succeed. A warm state alone does not mean the instance is ready for users.
The hot message promotes a warm instance; it does not convert an active instance back to warm mode. To re-establish a warm standby, stop the former active services through the applicable recovery procedure, restore the appropriate database and file configuration, and start that standby environment again with -warm.