Serviceguard Manager Version A.05.02 Release Notes, Third Edition, July 2009

Service Command = /usr/bin/X11/xclock -title "My Xclock"
When this is entered into the Service Command field and applied, the target control script in
/etc/cmcluster/<pkgName>/<pkgName>.sdf.sh would be found to contain:
SERVICE_CMD[0]="/usr/bin/X11/xclock -title My Xclock"
This command does not function, because if you try to invoke this, xclock would give a usage
error.
What is the workaround? There is no workaround.
JAGaf14625 Problems creating and modifying packages that contain non-alphanumeric characters
(,*,$,#, ...).
What was the problem? The problem affects the configuration (creation or modification) of packages
on HP-UX.
Serviceguard allows a package to contain any of the following special characters in the package
name:
x+ = ` ' { } [ ] | ; " < > ? # ( )
But if the user enters a package name containing one of these special characters, an error will be
generated by the Cluster Object Manager when the configuration is applied.
These characters are acceptable in the package name: _ - :
What is the workaround? There is no workaround.
JAGaf43912 Cluster Properties: wrong lock info in older version of COM
What was the problem? Versions of COM older than B.03.01 may display wrong information about
the HP-UX cluster lock, because it may not support discovery of a Lock LUN. The COM (Cluster
Object Manager) is a component of Serviceguard.
What is the workaround? If you are using Lock LUN in a Linux cluster, upgrade COM to version
B.03.01 or later. The COM (Cluster Object Manager) is a component of Serviceguard. For more
information, see the documentation at http://docs.hp.com/hpux/ha -> Serviceguard.
JAGaf64743 Help: Problem entering Locale ko chars in index of online help viewer
What is the problem? Entering ko (Korean) characters into the index of online help can cause the
help window to hang.
What is the workaround? There is no workaround.
JAGaf48460 Event browser displays wrong status of snmp subagent
What is the problem? The event browser may incorrectly say that the SNMP subagent is down,
and the trap is not set, even when they are active and reporting new events.
What is the workaround? Follow these steps:
1. Bring up Set Trap dialog for the managed node.
2. Type in something in the textfield to force trap setting.
The status should now be Trap Set.
The back-end logic is:
Step 1: Use the community name supplied to set trap (Success or Failure)
Step 2: Query the status of the trap destination. Since the trap destination is already set in the
backend, Step 1 will not unset the trap, even if the community name is not correct. Step 2 will
then display the updated status.
Note: If the status is already Trap Set, unsetting the trap destination in the backend will not be
visible in the front-end GUI. In this case, please disconnect and then reconnect. Serviceguard
Manager uses both snmp event notification and polling to ensure the most up-to-date data is
42 Serviceguard Manager Version A.05.02 Release Notes