A.1 Overview of Error Messages

Errors that deal with command usage and other simple error messages are not documented in this appendix. Examples and explanations of these messages are:

command name: flag unknown flag

The above message indicates that an unsupported option (flag) was used with the Solstice HA command, command name.

command name: command line error

The above message indicates that a command-line error was detected when the Solstice HA command, command name, was invoked.

command name: must be root to run this command

The above message indicates that the user must be superuser (root) to invoke the Solstice HA command, command name.

A.2 Membership Monitor Messages

The following messages are returned by the membership monitor daemon. All of the errors result in the daemon calling the abort programs and a takeover being performed by the sibling.

add_hostname: nodeid id is out of range, nodeid

The nodeid specified in the cmm_confcdb file has a value greater than 32. If this message is displayed, contact your service representative.

comm_addnode: duplicate nodeid id, nodeid

You should contact your service provider if this message is displayed.

newipaddr: unknown host hostname

The cluster monitor is unable to obtain information about the private network names specified in the cmm_confcdb file. A possible problem is that the private network names specified are not in the /etc/hosts. Add the private network names to the /etc/hosts file. This problem should have been discovered by either hasetup(1M) or hacheck(1M) during initial configuration.

node cannot bind to any host address

The cluster monitor is unable to bind to any of the addresses specified in the cmm_confcdb file. These addresses would be the names mapping to the private Ethernets. The problem could be that the interfaces are not configured or that the cable connections are wrong. You should check the cable connections and test the connections using ping(1M). This problem should have been discovered by either hasetup or hacheck during initial configuration.

t_bind cannot bind to requested address

The port number specified in cmm_confcdb is already in use. To correct the problem, specify a new port number. Both servers should be using the same number. Specify the new port number using the hasetup command.

nodetimeout cannot be lower than msgtimeout

The default values in the cmm_confcdb file are corrupted. If this message is displayed, contact your service representative.

cdbmatch value, fullkey, cdb_sperrno

The values in the cmm_confcdb file are corrupted. If this message is displayed, contact your service representative.

invalid value for parameter failfast

The fastfailmode parameter in the cmm_confcdb file is wrong. If this message is displayed, contact your service representative.

parameter parameter must be an integer

A parameter in the cmm_confcdb file is corrupted. If this message is displayed, contact your service representative.

parameter parameter must be either true or false

A parameter in the cmm_confcdb file is corrupted. If this message is displayed, contact your service representative.

parameter parameter must be an numeric

A parameter in the cmm_confcdb file is corrupted. If this message is displayed, contact your service representative.

parameter parameter not found in config file

A parameter is missing from the cmm_confcdb file. If this message is displayed, contact your service representative.

unsupported version number number (supported number)

The version number of the cmm_confcdb does not match the version number expected by the cluster monitor. The default file has a value of 2. Install the same Solstice HA packages on both systems.

must be superuse to start

A user without superuser privileges attempted to invoke the cluster monitor. This only occurs if the user attempted to start the cluster monitor from the command line. Either allow the monitor to start automatically when the system is rebooted or run the following command:

# /etc/init.d/SUNWhadf start

Aborting node with stale seqnum number

A server is in an unexpected state.

received signal signal

The cluster monitor, clustd, received either a SIGTERM, SIGPOLL, SIGALRM, or SIGINT signal.

transition timeout

The cluster reconfiguration program was not completed within the specified transition timeout. This error will result in a takeover being performed by the other system in the configuration.

unknown scheduling class class

An invalid scheduling class was specified in the cmm_confcdb file. The only valid choice is RT. To correct the error, restore the template cmm_confcdb file and enter a valid class using hasetup.