Attribute |
Value |
Description |
AddUnrecognizedNodes |
|
Specifies if the CSM management server should accept unexpected
requests from systems, to manage them and allow them to join the CSM cluster.
It is only necessary to set this to 1 if for some reason you need
to initiate a join from the node (using the mgmtsvr command directly)
instead of from the management server (using installnode or updatenode). |
ClusterSNum |
|
Sets the cluster serial number to this string. This value is used for service
and is unrelated to the serial numbers of any of the hardware. |
ClusterTM |
|
Sets the cluster type and model to this string. This value is used for
service, and should be in the form ####-###. |
ExpDate |
$values{ExpDate} |
The expiration date for the current try-and-buy license key. This value will be blank if a full license has been installed. This attribute cannot be set. Use the csmconfig –L command to install a full license key. |
HeartbeatFrequency |
|
Number of seconds between heartbeat messages sent to a node. The
default is 12 seconds. Along with the default of 8 for HeartbeatSensitivity,
this means that it will be at least 96 seconds after a node goes down before
the CSM node Status attribute reflects that fact. The reason the
default is so long is to prevent false downs caused by network traffic
or node load. If you want the Status attribute to be more responsive,
lower either of these values. Lowering HeartbeatFrequency will slightly
increase network traffic. |
HeartbeatSensitivity |
|
Number of missed heartbeat messages sent to a node before that node is
declared unreachable. The default is 8 times. Lowering this
value will make the node Status attribute more responsive, but runs the
risk of an occasional false down. |
MaxNumNodesInDomain |
$values{MaxNumNodesInDomain} |
Specifies the maximum number of nodes allowed in this CSM cluster. This could be affected by the license key that is installed using the csmconfig –L command, but currently all licenses are unlimited. |
PowerStatusMode |
|
Sets the mode the HWCTRLRM uses to retrieve and maintain the
PowerStatus attribute in the ManagedNode table
for all of the nodes. Valid values are:
- mixed - either poll or use events depending on the best method for each hardware
control point (HCP) type.
- polling - periodically query the power status of each node. Since each query is logged in the RSA/MM log, this can cause the log to fill up eventually. The frequency of polling is an attribute of the IBM.HwCtrlPoint resource class. It can be display using lsrsrc -t IBM.HwCtrlPoint . It can be changed for all of the HCPs of a particular type
using chrsrc -s "PowerMethod='xseries' " IBM.HwCtrlPoint PollingInterval=400 , although this value will not be remembered across restarts of the HWCTRLRM.
- events - use only events to have the HWCTRLRM notified when the power status
of a node changes. Currently, the events from all types of HCPs are
not completely reliable.
- none - do not maintain the PowerStatus attribute. The rpower -n nodes query will still return the correct values.
|
RegSyncDelay |
|
Delay in seconds between the time the cluster and node data is
updated in memory and when it is written onto the disk. Increase
this value to improve the performance of definenode, chnode, rmnode, and nodegrp in large clusters. |
RemoteShell |
|
Specifies the path of the remote shell command that CSM and dsh should
use to run commands on the nodes. The default is /usr/bin/ssh. |
SetupRemoteShell |
|
Indicates whether CSM should try to automatically set up security for the
remote shell specified in the RemoteShell cluster attribute. Currently
this is only supported for /usr/bin/ssh and /usr/bin/rsh. (In order
for rsh set up to be successful, rexec must already be enabled on each
of the nodes.) |