4.28. SCSI Persistent Reservations
fence_scsi, the fence agent for SCSI persistent reservations.
- When using SCSI fencing, all nodes in the cluster must register with the same devices so that each node can remove another node's registration key from all the devices it is registered with.
- Devices used for the cluster volumes should be a complete LUN, not partitions. SCSI persistent reservations work on an entire LUN, meaning that access is controlled to each LUN, not individual partitions.
/dev/disk/by-id/xxx where possible. Devices specified in this format are consistent among all nodes and will point to the same disk, unlike devices specified in a format such as
/dev/sdawhich can point to different disks from machine to machine and across reboots.
Table 4.29. SCSI Reservation Fencing
|luci Field|| ||Description|
|Name|| ||A name for the SCSI fence device.|
|Unfencing|| || When enabled, this ensures that a fenced node is not re-enabled until the node has been rebooted. This is necessary for non-power fence methods (that is, SAN/storage fencing). When you configure a device that requires unfencing, the cluster must first be stopped and the full configuration including devices and unfencing must be added before the cluster is started. For more information about unfencing a node, see the |
|Node name|| ||The node name is used to generate the key value used for the current operation.|
|Key for current action|| ||(overrides node name) Key to use for the current operation. This key should be unique to a node. For the "on" action, the key specifies the key use to register the local node. For the "off" action,this key specifies the key to be removed from the device(s).|
|Delay (optional)|| ||The number of seconds to wait before fencing is started. The default value is 0.|
Figure 4.21. SCSI Fencing
ccs -f cluster.conf --addfencedev scsifencetest1 agent=fence_scsi
cluster.confentry for the
<fencedevices> <<fencedevice agent="fence_scsi" name="scsifencetest1"/> </fencedevices>