samd(1M)


samd -- System Availability Monitor daemon

Synopsis

/var/avail/sam/samd CLUS_FAIL | NODE_FAIL | PROC_FAIL

/var/avail/sam/samd -h

Description

Do not use samd on the command line, except with the -h option.

The samd daemon starts a pair of persistent processes that watch for cluster failures, node failures, their own failures, and significant time changes. The main process is samd and is monitored by the keepalive system. The second process is samsigd, and is started and monitored by samd. It catches node event signals and passes this information to samd for processing. Both processes are always operational.

init(1M) and keepalive(1M) normally start and stop samd. Should you need to manually stop or restart samd, use the samdctl(1M) command.

If the cluster is rebooted while samd is manually stopped or down for other reasons, the down event of the cluster is time stamped as approximately when samd stopped running. Thus, the duration of the cluster failure is overstated in this case. Do not manually stop this daemon before shutting down the system.

The samd daemon logs information to the event log file using the samlog command.

Two daemon parameters can be modified by a user with appropriate permission, although normally the default values do not need to be changed. 

Before modifying these values, consider the consequences of the change you plan to make. For example, if the default time-change window is set too low, an excessive number of time-change events may be logged, making samrep reports cumbersome. 

To change these defaults, you must edit the samd script. Use the samdctl stop command to stop the daemon, modify the parameters in the /var/avail/sam/samd file, then use samdctl restart to restart the daemon. The modifiable parameters are:

HEARTBEAT_SEC_DEF
Sets the number of seconds the daemon periodically sleeps. The default is 10 seconds. This parameter affects the accuracy of event timestamps to within the number of seconds specified. The number of seconds cannot exceed 30, which is the maximum allowed. You may want to change this value to alter the CPU utilization of the samd daemon or to adjust the accuracy of the timestamps.
TIME_CHG_WINDOW_SEC_DEF
Controls the amount of time discrepancy allowed before a time-change event is recorded in the event log file. The default is 300 seconds. By default, time discrepancies of less that five minutes are not recorded. You may need to change this value when the system time is frequently reset by a process such as timed(1Mtcp).

Modifications to the samd script are not preserved across subsequent installations of SAM. Additionally, modifications to this script cause subsequent pkgchk sam commands to fail.

Options

The samd daemon runs with one of the following arguments:
CLUS_FAIL
This startup occurs after a failure of the cluster.
NODE_FAIL
This startup occurs after a failure of the node on which samd is running.
PROC_FAIL
This startup occurs after a failure of the samd process itself.

Files

/var/avail/sam/samd
samd daemon script file
/var/avail/sam/eventlog
SAM event log file
/var/avail/sam/.state
SAM private state file directory for lock and state files

References

intro_sam(1M), samdctl(1M), samlog(1M), samrep(1M)
03 Feb 2000
© 2000 The Santa Cruz Operation, Inc. All rights reserved.
UnixWare 7 Release 7.1.1b - 14 April 2000
© 2000 Compaq Computer Corporation.