Memo for the Record                               15 June 1981

Subject: Setting up the management reporting system

1.  _P_r_o_b_e This command resides in /sys/prog/probe and should be
executed once when the system is rebooted.  Probe should be "&"ed
off with nohup by whatever command "brings up" your system.  The
shell files "operate.sh" and "leton.sh" are provided as examples
of how the AFDSC systems are brought up.  The command:

 nohup /sys/prog/probe -f -c 180&

will generate probe records to the file (-f) /usr/adm/pr.data
every 180 seconds (-c 180).  When the system is available for
users, give the command:

 /sys/prog/probe -f -u

which will signal the running probe that the system is UP (-u).
When the system is brought down key:

 probe -f -d

to signal the running probe to generate a down (-d) record.  The
shell files "drain.sh" and "kickoff.sh" are provided as examples
of how the AFDSC manages bring the system down.  To Kill off
probe cleanly key:

 probe -f -k

which will kill probe off so it doesn't generate a garbage
record.  At midnight 0000 hours probe does some End of Day pro-
cessing where it copies the /usr/adm/pr.data file to
/usr/adm/pr.hold.   If your site is not up a midnight, the "eod"
subroutine in probe should be changed to trigger at a different
time.  Probe should be in execution at all times even if the sys-
tem is not available for users to properly gather its data.  At
the AFDSC all aspects of probe processing are managed by the
shell files "operate.sh", "kickoff.sh", "drain.sh", and
"leton.sh" so the operations personnel are never concerned with
the actual "probe" commands themselves.

2.  _M_g_t._s_u_m Mgt.sum reduces the holding files /usr/adm/sh2.hold
(shell accounting), /usr/adm/log.hold (login accounting), and
pr.hold (probe accounting).  Since the holding files are created
about midnight mgt.sum can be run anytime after holding file
creation (by the shell file /sys/prog/acct).  At the AFDSC
mgt.sum is run at about 0630 (via cron) in the morning to avoid
dumps and prime user time.  The output of mgt.sum is the direct
access database /usr/adm/mgt.data which stores one year's worth
of reduced daily data.  The shell file "mrpt.sh" is provided as
















an example of how the AFDSC accomplishes both the mgt.sum and
mgt.rpt function each day.

3.  _M_g_t._r_p_t Mgt.rpt provides a formatted report in two formats
from the /usr/adm/mgt.data database see the man page for further
details.  The shell file "mrpt.sh" is provided as an example of
how the AFDSC accomplished the mgt.sum and mgt.rpt function each
day.

4.  _A_c_c_t._s_h This shell file should be run just before midnight
(1150 pm will do) to copy the /usr/adm/log.act (login accounting)
file to /usr/adm/hold, and /usr/adm/sh2.act (shell accounting) to
/usr/adm/sh2.hold.  The shell file ensures that the copies will
be "disconnected" from the original files which may still be open
by any user's shell currently logged in at the time.

5.  _M_r_p_t._s_h This shell file runs the "mgt.sum" program and
"mgt.rpt" command to generate a daily report.  Run this through
"cron" sometime after midnight.


































2











