| Hints and Tips |
This section provides advice and guidance on configuring RSF-1 and building highly available data services to be monitored, controlled and failed over inside the cluster framework.
|
| Filesystem Mounting |
Remember to set any data service's clustered filesystems to not automatically mount via the "mount at boot" option when the node is booted. RSF-1 itself will automatically check and mount any clustered filesystems as part of the service's startup sequence.
|
| Faster Failover |
Dramatically decrease failover times simply by enabling journaling for each service's clustered filesystems, provided the target platform supports journaled filesystems.
|
| Name Services |
When configuring an RSF-1 cluster, it's a good idea to add all cluster addresses, i.e. network heartbeat interfaces and service virtual hostnames, to the "/etc/hosts" file on all cluster nodes and set "/etc/nsswitch.conf" to reference it first, as DNS outages can interupt correct cluster operation and prevent services from failing over.
|
| Stop Scripts |
Save time when building data services, instead of generating each service's stop scripts manually, simply use the "rsfklink" command to generate the "K" scripts automatically inside each service's runtime configuration directory.
|
| Service Timeouts |
When building data services, it's a good idea to set the RUNTIMEOUT parameter to be at least 60 seconds for production services and the INITIMEOUT parameter to be roughly one and a half times the time it takes in seconds for the service's primary server to boot normally.
|
| Configuration File |
When configuring RSF-1, the configuration file ("/opt/HAC/RSF-1/etc/config") has to be identical on all cluster nodes. It's a good idea to create the configuration file on one node first, then simply remote copy (or FTP) it over to the remaining node(s) in the cluster.
|
|
