This is a simple shell script that issues a failover[1] from the node[2] where it is running to the other node, if available. It is useful to interface with an external resource monitor, such as mon[3] or Nagios[4], or any home-made-monitor-script.
hb_standby is a script that allows manual control of which node is controlling what resources[5] at a given time. When the "standby" option is provided to the Heartbeat init script, it calls hb_standby with the "all" option.
usage: hb_standby [all|foreign|local|failback]
The full path is usually /usr/lib/heartbeat/hb_standby. To get a list of the options to hb_standby issue the command, "/usr/lib/heartbeat/hb_standby --help". NOTE: For historical reasons, unlike most scripts where failure to specify an option gives you the list of possible options, failure to provide an option to hb_standby is equivalent to "hb_standby all".
The meaning of the options is relative. For this sample haresources[6] file (IMPORTANT: all resources for one node on one line):
nodea drbddisk::r0 Filesystem::/dev/drbd0::/local/groups::ext3 10.0.0.1 smb nodeb drbddisk::r1 Filesystem::/dev/drbd1::/local/ldap::ext3 10.0.0.2 ldap
When run on nodea, means migrate any resources that nodea is the preferred node[7] for. In this case Samba would be shutdown, the IP address 10.0.0.1 would be released, /local/groups would be unmounted, /dev/drbd0 would be placed into secondary and nodeb would take all these services over.
When run on nodeb, OpenLDAP would shutdown, 10.0.0.2 would be released, /local/ldap would be unmounted, /dev/drbd1 would be placed into Secondary mode and nodea would take over all these services.
When run on nodea, means migrate any resources that nodea is NOT the preferred node[7] for. In this case OpenLDAP would shutdown, 10.0.0.2 would be released, /local/ldap would be unmounted, /dev/drbd1 would be placed into Secondary mode and nodeb would take over all these services.
When run on nodeb, Samba would be shutdown, the IP address 10.0.0.1 would be released, /local/groups would be unmounted, /dev/drbd0 would be placed into secondary and nodea would take all these services over.
Regardless of which node it's run on means to release all resources so the other node takes over all resources.
Without an argument is the same as "hb_standby all"
Same as "hb_standby foreign"
haresources[6], hb_takeover[8], cl_status[9], mon[3], failback[10]
| [1] | http://en.wikipedia.org/wiki/Failover |
| [2] | http://www.linux-ha.org/ClusterNode |
| [3] | http://www.linux-ha.org/mon |
| [4] | http://en.wikipedia.org/wiki/Nagios |
| [5] | http://www.linux-ha.org/resource |
| [6] | http://www.linux-ha.org/haresources |
| [7] | http://www.linux-ha.org/PreferredNode |
| [8] | http://www.linux-ha.org/hb_takeover |
| [9] | http://www.linux-ha.org/cl_status |
| [10] | http://www.linux-ha.org/failback |
This information provided courtesy of the Linux-HA project at http://linux-ha.org/