ALARM_SERVER Aegis ALARM_SERVER
NAME
alarm_server - notify user about special events
SYNOPSIS
/etc/sys/alarm/alarm_server [options]
DESCRIPTION
The alarm server looks for a variety of conditions that may
be of interest to users or system administrators. When one
of those conditions arises, the alarm server pops a window
on your display describing the condition and sounds an
audible alarm.
You should normally run the alarm server in a background
process; this means starting it with the DM command cpo.
See EXAMPLES.
The conditions on which the alarm server can report are:
Potential disk overflow
You are notified when the disk containing
your / directory (diskless users please note)
starts to run out of free space.
Severe network problems
You are notified whenever there is a new
network hardware failure message, indicating
severe network problems. The hardware failure
message is described in the /com/netstat
documentation.
Observer reports from netmain_srvr
You are notified whenever one of the
observers in the netmain_srvr program makes a
report on unusual network conditions. This
option is interesting only to the network
system administrator.
Brief messages from other users
You can use the send_alarm program to direct
short messages to other users or nodes (see
"related topics", below).
Optional software
The alarm_server is also used with certain
optional software packages. The appropriate
software user's manuals give complete
instructions on the use of the alarm_server
with the optional software packages (see
"related topics", below).
Each condition is checked once every four minutes, or at
some other interval set by the -period option, but alarms
may not be posted every time the condition is checked. See
the description of each alarm to find out what that alarm's
scheduling policy is.
Use alarm_server only through the DM command cpo. Do not use
it as a server process (via cps). Many people start their
alarm servers in their "~user_data/startup_dm" command
files.
OPTIONS
Default options are indicated by "(D)."
-disk[_full] [nn]
Notifies you when the disk containing your /
directory is more than <nn> percent full.
This alarm does not run unless you specify
the -disk_full option. If you omit <nn>, it
is set to 95 (percent full). If you do not
clean off the disk, or if the disk-full
condition recurs, the alarm is posted again.
After notifying you twice, the alarm will not
be posted again for at least an hour.
-hw[_fail] Notifies you when some node detects network
hardware problems (as seen in the 'Last ring
hardware failure' section of a "/com/netstat
-l" report). Hardware failures are not
checked unless you specify this option. Only
new failure reports cause alarms.
-netmain [pathname ...]
Enables alarms from netmain observers. The
pathname[s], if specified, represent text
files containing lists of node that run
netmain_srvr. If no pathname is specified,
the file
~user_data/alarm_server.netmain_srvr_list is
used. The files should contain lists of node
names or hex node ID numbers, separated by
spaces or on different lines. Comments in
these files start with { or # characters, and
run to the end of the line.
The alarm server reads these files only when
it starts up. If you add or delete node
names in these files after the server is
running, your changes will not take effect
until the next time the server starts up.
-nm_srvr node_spec [...]
Enables alarms from netmain observers.
<node_spec> specifies a node from which you
want observer alarms. <node_spec> may be a
hex node ID number or a node name.
-nonetmain (D) This option prevents the alarmer from
checking for observer alarms from the
netmain_srvr program.
-msg (D) Allows the alarm server to receive messages
from the send_alarm program. The mbx_helper
program must be running for this to work.
-nomsg Prevents the alarm server from receiving
send_alarm messages.
-nets Notifies you when a network disappears from
your internet or appears in your internet.
This is useful only in Domain internet
environments.
-ndist Like -nets, but also notifies you of changes
in the number of hops to a distant network.
This is also useful only in Domain internet
environments.
-nonets (D) Suppresses alarms describing changes in the
internet topology.
-bell1 Normally, a distinctive tone pattern
accompanies each kind of alarm. If you
specify this option, a single short beep is
used for any alarm.
-nobell Normally, each kind of alarm is accompanied
by an audible alarm. This option suppresses
the audible alarm.
-p[eriod] nnn Each alarm-detector is checked every <nnn>
minutes. By default, nnn is 4 minutes. It
must always be at least one minute.
-v[ector] dx [dy]
Selects the offset from each alarm window to
the next, in pixels. Defaults: dx = 235, dy
= 0.
-w[indow] initx [inity [width [height]]]
Sets the screen position of the first alarm
window and the size of the alarm windows, in
pixels. Defaults: initx = 1, inity = 1,
width = 225, height = 100.
EXAMPLES
Command: cpo /sys/alarm/alarm_server -disk 98 -bell1
Command: cpo /sys/alarm/alarm_server '*~user_data/opts' -n alarms
This is a very useful form of the command. It lets you
create an options file (in this case ~user_data/opts)
and use that file to control the alarm server. The alarm
server's process is named by the -n option. In this case,
the process is called "alarms".
Note on acls:
Occasionally, the alarm server will print a warning message
about a file called '.../alarm_server.msg_mbx' as it starts
up. These messages can come from a variety of basic
problems, but many times the problem comes from acls. The
alarm server can usually handle these acl problems by
changing the acls on its message mailboxes automatically,
but it will sometimes need outside help.
If you want to receive send_alarm messages from other nodes,
you need to run the mbx_helper, and the acls on two files
must allow the mbx_helper program read/write access. You
should normally run mbx_helper via the DM command cps; this
means it is assigned an SID of
`node_data/alarm_server.msg_mbx
<your-home_directory>/user_data/alarm_server.msg_mbx
both allow read/write access to 'user.server.none'.