sysctlrd - communicates with the system controller and LCD front panel on
Onyx/Challenge L/XL systems
/usr/etc/sysctlrd -d [ -g ] [ -n ] [ -v ] | -p
sysctlrd can be run as a user command or as a daemon run by
/etc/rc2.d/S33sysctlr. It handles all necessary communications with an
Onyx or Challenge L/XL system's system controller, which controls power
sequencing and fan speed, monitors system sensors, and drives the LCD
front panel. sysctlrd runs the cpu activity histogram on the front
panel, retrieves environmental sensor data, and handles warning and alarm
messages from the system controller. sysctlrd also logs configuration
error messages from the PROM in /var/adm/SYSLOG (see syslog(3C)).
The options are:
-d Daemon mode. Only one sysctlrd can run as a daemon at a time, and
it must be run as root.
-g Graceful powerdown mode (only works with -d). sysctlrd sends a
hang-up signal to all processes, waits two seconds, sends a kill
signal to all processes, syncs all disks, and then powers the system
-n No CPU meter mode (only works with -d). Keeps sysctlrd from sending
CPU performance information to the LCD front panel.
-v Give more verbose output.
-l Logging of VDC extrema changes from all of the power supplies when
run in VDC variance logging mode. (e.g. the "-l" option is passed to
sysctlrd from /etc/init.d/sysctlr). This is useful if you are
collecting lots of information to debug and/or analyze a problem
which you believe may be related to power surges and/or hardware
failures. Please contact SGI technical support in the event that
there is some problem which you believe needs their attention.
-p Print sensor information (does not work with -d). This option
causes sysctlrd to query the daemon for system controller sensor
information and print it in a human-readable format.
"*** ..." messages are PROM configuration messages logged by sysctlrd.
Messages of the form "Event: date time: ..." are retrieved directly from
system controller log RAM. These may hold information about when and why
the system was shut down previously. Note that the times listed are from
the system controller's internal clock. This clock is set at each
invocation of sysctlrd with the -d option, and it runs freely from then
on. The clock has no concept of alternate timezones. Some examples of
informational messages that may be logged are:
SCLR DETECTED The system was reset from the front panel or rebooted via
a UNIX command.
SYSTEM OFF The system was powered-off from the front panel.
SYSTEM ON The system was powered-on from the front panel.
NMI Someone selected NMI (nonmaskable interrupt) from the
"Couldn't open /dev/sysctlr. Exiting" is printed to standard error if the
daemon cannot open the system controller device driver. It usually means
that a daemon is already running.
"Get env info failed" is logged if sysctlrd is unable to get the sensor
information after retries.
"Get scale failed! Assuming small panel" is logged if sysctlrd cannot
get the LCD panel size. On Onyx/Challenge XL systems, this results in
very short CPU performance bars since they are improperly scaled.
"Get log failed!" is logged if sysctlrd cannot retrieve the system
controller event log containing such events as system resets, power
fails, voltage problems, overheating, and so on.
"Can't contact daemon. Exiting" is printed to standard error if a nondaemon
invocation of sysctlrd cannot contact the daemon to retrieve
information. The daemon may have died or may not have been able to set
up its IPC.
"Giving up on fetching environmental information" is logged if sysctlrd
fails to retrieve environment information 20 times in succession. Nondaemon
invocations cannot retrieve information after this happens.
"Overtemp alarm!" is printed to standard error when the system controller
sends an overtemp message. This system is shut down automatically at
this point, syncing disks if graceful powerdown is active.
"Keyswitch off!" is printed to standard error when the key is switched
off with sysctlrd running. This system is shut down, syncing disks if
graceful powerdown is active.
"Blower failure!" is printed to standard error if the system controller
detects a failure in one of the system's fans. The system is powereddown
"Voltage out of tolerance!" is logged if the system controller detects a
voltage that is out of tolerance but not yet dangerous to the system.
"Firmware compensating for blower RPM problem" is logged if the system
controller must compensate for a blower speed problem by requesting a
higher blower speed.
"System controller firmware reset" is logged if the system controller
firmware detects an internal error and resets itself. This message is
only indicative of a problem if it is logged repeatedly.
"COP timer reset error!" is logged if the system controller internal
timer detects hung firmware. Again, this message is only indicative of a
problem if it is logged repeatedly.
"System controller crystal oscillator failed!" is logged if the system
controller detects a problem with its own oscillator.
The system controller will send a warning message when it measures an RPM
higher than the upper tolerance. Normally this message is sent
immediately after the system controller powers up (this is only for the
systems that have the 2700 RPM blower installed). Sometimes, the blower
sends out bad RPM data. The measurement data shows RPM values > 28000
RPM. In cases such as these, it has been determined that the tach pulse
from the blower has noise on it, which causes these incorrect high
readings. Some blower(s) checked were actually spinning near 1400 RPM
when they were reporting an RPM value of 27K. Now, the system controller
monitors for this condition and if a reading is posted by the system
controller > 3000 RPM, a notification message will be posted as a warning
in both the SYSLOG and on the console at the first occurrence of such an
event and at intervals of one hour thereafter (if the condition
persists). Please contact SGI technical support in the event that you
constantly receive such a message from the system controller.
Note: running the system controller daemon with verbose mode on (-v) will
generate a large amount of output in the system logfile and should only
be used when debugging hardware configurations.
PPPPaaaaggggeeee 3333 [ Back ]