8.0;netmain_srvr (network_maintenance_server), revision 8.0, 84/02/18
NETMAIN_SRVR (NETWORK_MAINTENANCE_SERVER) -- Collect network error stats.
usage: NETMAIN_SRVR [-L[OG] [pathname] | -NL[OG]] [-LL n]
[-S[AMPLE] probe time ...]
[-A[PPEND]|-NA[PPEND]
[-SK[IP] probe distance ...]
[-TOPO[_INIT] pathname] [-CMDF pathname]
[-OBS[ERVE] observer time ...]
[-RE_OBS[ERVE] observer time ...]
[-NTOPO[_INIT]]
FORMAT
NETMAIN_SRVR [options]
NETMAIN_SRVR collects and stores performance statistics for the network. You
run it as a server process (via the DM command CPS), to gather information to
analyze with the NETMAIN program. See Administering Your DOMAIN System for
complete details. This help file describes parameters for NETMAIN_SRVR that
you can set from the command line when you start the monitor. Note that you
can change any of these in the NETMAIN environment, once the server is
running.
NETMAIN_SRVR has a number of options. Instead of including them all on the
command line you can use an options file, by specifying the "-CMDF" option.
If you specify "-CMDF pathname", the server first reads the options listed in
the options file specified, and then reads any other options on the
NETMAIN_SRVR command line. If there are any conflicts between the options
file and the command line, the command line settings are used. For example, if
the options file specifies -LL 1500 and the command line specifies -LL 3000,
3000 is the limit on the log file's length.
If a NETMAIN_SRVR does not start properly, a record of the failure appears in
`NODE_DATA/NETMAIN_SRVR.ERR_LOG. See Administering Your DOMAIN System for
more information.
OPTIONS
Default options are indicated by "(D)."
-APPEND Append to an existing log file with this name, if one
already exists, otherwise create a log file with this name.
This option is only valid when a log file pathname is
specified with the -L option. Contrast this with the
-NAPPEND option.
-CMDF pathname
Accept options from an ASCII text file 'pathname'. You may
use this option only from the command line, not in the
options file. There can only be one options file.
-L[OG] [pathname] (D)
Create a log file. Optionally, specify a pathname, which is
relative to the `NODE_DATA/NET_LOG directory. If either
this option or the pathname is not specified, the log file
name is derived from the current date:
`NODE_DATA/NET_LOG/NET_LOG.yy.mm.dd. The log file is
stored on the disk of the node running NETMAIN_SRVR, and
must remain there for NETMAIN_SRVR to write to it.
-LL n (D) Set an upper limit on the length of the log file. The file
size limit (n) is in 1024-byte units. The default value for
'n' is 3000. You must use this option when you start the
monitor if you don't want to use the default length for the
first log file, since you cannot change the name of a log
file once it's open.
-NAPPEND (D) Create a new log file, over-writing any log with that name,
if one exists. This option is only valid when a log file
pathname is specified with the -L option. Contrast this
with the -APPEND option.
-NL[LOG] Do not write to a log file. NETMAIN_SRVR still runs probes
and observers.
-NTOPO[_INIT] (D)
Override the -TOPO_INIT option, if -TOPO_INIT is specified
in an options file. -NTOPO is useful only on the command
line.
-OBS[ERVE] observer time ...
Set the interval at which the named observer wakes up.
Specify 'time' as hh:mm:ss, hh:mm, or 'never', if you do
not want the monitor to run the observer. Multiple
observer/time pairs are permitted. See the default times
listed below for each observer.
-RE_OBS[ERVE] observer time ...
Set the "Recheck interval" -- the interval that the
observer waits before rechecking a node that has caused an
alarm condition. By setting a recheck interval you insure
that the observer only reports on a node once every 'time'
period. If the recheck interval is too short, the observer
may produce many redundant alarms. Specify 'time' as
hh:mm:ss, or hh:mm. Multiple observer/time pairs are
permitted. See the default recheck intervals listed below
for each observer.
-S[AMPLE] probe time ...
Set the interval at which the named probe wakes up.
Specify 'time' as hh:mm:ss, hh:mm, or 'Never', if you do
not want the monitor to run the probe. Multiple probe/time
pairs are permitted. See the default times listed below
for each probe.
-SK[IP] probe distance ...
Set the skip distance for the probe named. If the skip
distance is n, the named probe samples approximately 1/n of
the nodes every time it wakes up. Multiple probe/distance
pairs are permitted. See the default skip distances listed
below for each probe.
-TOPO[_INIT] pathname
Initialize the monitor's total node list from a data file.
The file may contain any number of node names or
hexadecimal IDs, separated by spaces or on separate lines.
If there is a '#' or '{' in any line, that character and
all characters to the right of it are ignored (i.e., # and
{ are comment markers).
The monitor contains these probes:
Default
sample Default
Name time skip Data collected
============ ======= ======= ===================================
CPU_TIME 0:30:00 1 Null/AEGIS/user CPU time
DISK_ERRS 0:30:00 1 Disk and storage module errors
ERR_COUNTS 0:30:00 1 Network error counts (normal traffic)
EST_TOPOLOGY 1:00:00 NA Total LCNODE estimate
HW_FAIL 0:01:00 NA Hardware failure messages
NET_SERVICE 0:30:00 1 Network service queue stats
PAGING 0:30:00 1 Diskless/partner info.
SWD_10_MSGS Never NA Network error counts (10 special msgs)
SWD_100_MSGS Never NA Network error counts (100 special msgs)
TIME_SKEW 3:00:00 1 Difference between node clocks
TOPOLOGY 1:00:00 NA Updates total LCNODE estimate
The monitor also contains these observers:
Default Default
Name interval recheck Situations reported
============ ======== ======== ===================================
MODEM_ERRS 0:30:00 12:00:00 Reports on nodes showing more than
five times the average number of
transmit modem errors.
WIN_CRC 0:30:00 12:00:00 Reports on winchester drives showing
excessive CRC errors.
The interactive help in NETMAIN lists the probe that collects each performance
statistic.
EXAMPLES
1. Command: CPS /SYS/NET/NETMAIN_SRVR -LL 1500 -L tuesday_error_log
2. Command: CPS /SYS/NET/NETMAIN_SRVR -S error_counts 0:15 hw_fail never
3. Command: CPS /SYS/NET/NETMAIN_SRVR -CMDF /sys/net/start.net_srvr -LL 3000
{ The file /sys/net/start.net_srvr might contain }
{ these lines: }
{ -L -LL 1500 }
{ -SAMPLE err_counts 0:01:00 -SKIP err_counts 30 }
{ -SAMPLE topology 0:20:00 }
{ -SAMPLE disk_errs 0:01:00 -SKIP disk_errs 30 }
{ -SAMPLE time_skew never }
{ -OBSERVE modem_errs 0:10:00 }
RELATED TOPICS
More information is available. Type:
- HELP CPS
for information about running server processes.
- HELP NETMAIN
for instructions on controlling NETMAIN_SRVR after it starts, and on
analyzing the data collected.
- HELP NETMAIN_NOTE
for information about adding notes to the network error log.
- HELP NETMAIN_CHKLOG
for information about detecting and deleting corrupt log files.