 | |||||||||||
| |||||||||||
 |
|||||||||||
| |
| ||||||||||||||||
| ShaoLin HA Cluster 1.0 Installation Manual and Operations Guide | ||
|---|---|---|
| Prev | Chapter 5. Configuring the clusters with Cluster Configurator Webmin interface | Next |
You can modify global options for the Cluster Manager . After modification of configuration settings, you will have to restart or reinitialize the clusters to take effect.
When the cluster boot up, it will wait a period in seconds for detecting if another cluster is alive or not. This value should be at least twice of the maximum timeout value among all heartbeat channels.
The default value for the keep alive period of heartbeat channels. The unit is in milli seconds. The default value is 3000 milli seconds. Notice, if individual keep alive period is specified, will override the default value.
The default value for the timeout period of heartbeat channel. The unit is in seconds. The default value is 10 seconds.
The maximum sync period when the system failovers. The unit is in seconds. If this timeout reached when syncing the data to disk, it will be stopped, and forced to failover, so you should ensure this value is reasonable and is sufficient for the sync action to complete. The default value is 5 seconds. If your system is heavily loaded, you should use a higher value. You should test by typing the command sync when the system is busy to measure the required value.
The default gateway will be used for standby cluster. You can leave it blank if your standby cluster doesn't need to reach outside the local subnet.
The minimum alive channels, if less than number of alive channel is alive, system will failover. Your previously active cluster will not be able to failback until a minimum number of live channels exists. This ensure the heartbeat communication is available before all channel die. You can't set this value less than 2. And if you set it to zero, means the minimum alive channel are all channels. The default value is 2 channels.
Master server have higher priority to become active node when two clusters boot at the same time. When a master node boot up while other cluster is on-line, it will automatically failback to the master node. Master server is specified by the giving the MAC address (for example: 0050fc3e7bd8) of one of the NIC of master server. MAC address can be lookup by using the command ifconfig -a.
The method of reseting the cluster, either shutdown or reboot. Beware that some of the machine don't support the shutdown method, please test it if you want to use shutdown method. If you have installed power switch, the cluster power is controlled by the power switch instead of software shutdown.
When two clusters boot at the same time, Cluster Manager will need to determine which server boot to boot as active and as well as standby. If set to ascending, the cluster with the smaller MAC address will boot to active. This behavior will be overwritten by the Master Server parameter.
Remote monitoring allow the Cluster Manager connect to view and control the cluster from a remote address using IP networks. To enable remote monitoring, set this to 'Yes'. It allow the Cluster Manager to connect to the cluster remotely, or else you can only connect to it locally. You will need this option set to 'Yes' if you want to connect to the standby cluster from other places (including from the active cluster and the web-based Cluster Manager UI).
The SMTP server for sending out email notifications. You must set a valid SMTP server if you want to receive events by email.
The sender name and email address of the email sending out by the HA Cluster. This is technically not useful but is used for the reader of the notification message to identify the source of this message.
The recipient email address of the email message sending out by the HA Cluster. If not specify, email notification will be disabled.
The IP address of the SNMP server that the SNMP traps will be sent to. If nothing specify, SNMP traps will be disabled.
Extra kernel modules you want to load in initrd.
The directory where the log file will be put. Default is set to /var/log/shaolin-ha.
The number of copies of log file will be keep, default is 4.
To restrict what kind of log message should be logged. There are seven log levels, EMERG, ALERT, CRIT, ERR, WARNING, NOTICE, INFO and DEBUG, say if you set the log level to ERR, then the EMERG, ALERT, CRIT and ERR log messages will be logged, whereas the WARNING, NOTICE, INFO and DEBUG log messages will not be logged. The default log level is INFO.
Log messages below or equal to this level will be notified by sending email or SNMP trap if SNMP server name is given and configured.
Log messages below or equal to this level will be notified by sending email or SNMP trap if SNMP is configured.
