Invokes the TCP Monitoring utility for capturing TCP related statistics. You may show TCP statistics (similar to the output of 'netstat -p tcp') and INET statistics, which displays kernel-level performance data. The /SHOW paramter determines which statistics are displayed. TCPMON is capable of producing CSV files which are compatible with T4, (Tabular Timeline Tracking Tool). Monitoring display may be affected by the following control keys: <CTRL-C> - exit <CTRL-Z> - exit <CTRL-Y> - exit <CTRL-T> - display a row of data <CTRL-W> - flush fileio Format: TCPMON
1 – Command Qualifiers
1.1 /BEGIN
/BEGIN=<time> Used to delay the start of monitoring. The time is entered as a VMS absolute time. E.g. 22:13 would start the monitoring at 10:13pm today. If it is not specified the monitoring begins immediately.
1.2 /CSV
/CSV=<filespec> /NOCSV (D) If specified, writes data to the CSV filespec. Data is written in a format which is compatible with the T4 utility or importing into a spreadsheet for further analysis..
1.3 /DISPLAY
If specified, writes data to SYS$OUTPUT:. If neither /CSV or /DISPLAY qualifiers are present on the command line, then /DISPLAY is the default. If the /CSV qualifier is given, then use the /DISPLAY qualifier to simultaneously write data to the CSV file and display the data to SYS$OUTPUT.
1.4 /END
/END=<time> Used to end the monitoring. The time is entered as a VMS absolute time. E.g. 22:13 would stop the monitoring at 10:13pm today.
1.5 /SAMPLE
/SAMPLE=<seconds> /SAMPLE=10 (D) Specifies the sampling interval. Snapshots of the statistics are captured at the end of the sample interval.
1.6 /SHOW
/SHOW[=(option[,...])]
/SHOW=TCP (D)
The following options can be selected:
ALL Print statistics for each of the registered subsytems.
[NO]INET Print kernel level statistics. The counters are:
Stats - if set, then profiling is enabled
PPE - if set, then the packet processing engine
is running.
%Idle - when PPE is running, this captures the
percentage CPU idle time
%Busy - when PPE is running, this captures the
percentage CPU busy time
Toggle - number of times PPE has toggled between
running and hibernating. Should be
fairly stable.
Move - Number of time PPE has moved between CPUs.
Conf - PPE reconfigurations across CPUs.
#Krp - TCP/IP kernel work queue contains kernel
request packets (KRPs). This statistice
represents the #Krps processed.
AvgQKrp - Average number of system cycles it takes
to queue each Krp on the KRP Queue.
AvDqKrp - Average number of system cycles it takes
to dequeue a KRP from the KRP Queue.
KrpQMax - Maximum length of the KRP queue.
KrpFrks - Not applicable to PPE. For the scalable
kernel this is the number of times a KRP
must be queued from one CPU to the TCP/IP
CPU via the system FORK_TO_CPU function.
This is an expensive operation.
[NO]TCP Print TCP statistics, similar to those available via
'netstat -p tcp'. The following statistcs are
displayed:
SPEED
-----
These include packet-rate and bit-rate for both
receive and transmit data.
TxPk/s - transmit data packet rate
TxMb/s - data rate in Mbits per second
RxPk/s - recieve packet rate
RxMb/s - data rate in Mbits per second
PACKET_SIZE
-----------
Average packet size over the sample period.
This is calcualted as data_packets/data_bytes
for both transmit and receive. The data is
derived from the RAW data:
TxPkSz - average number of bytes per packet sent
RxPkSz - average number of bytes per packet
received
CONNECTIONS
-----------
Connection statistics for both incoming and outgoing
connections. This displays the number of incoming and
outgoing connections. The data captured is:
In - number of connections accepted
Out - number of connections established
ERRORS
------
Not all statistics gathered here are necessarily
errors in the traditional sense. Many are protocol
related events which are bundled into this category.
The counters captured are:
RxDup - duplicate packets + packets with some
duplicated data.
RetxPk - data packets retransmitted.
RetxTo - retransmit timeouts
RetxDrp - connections dropped in retransmit
timeout
KpAlv - keepalive timeouts
KpDrp - connections dropped in keepalive
2 – Examples
$ set command tcpip$examples:tcpip$tcp_mon.cld $ tcpmon/samp=60/show=(all) This will monitor all subsystems with a sample interval of 60 seconds. $ tcpmon/show=(inet) This will monitor the INET subsystems with the default sample interval. $ tcpmon/csv=data_file.csv/output=data_file.txt/display This will monitor the default subsystems and write the data in CSV format to "data_file.csv" and the screen output will be written to "data_file.txt". Normally, no output will be displayed on the screen when it is being written to a file. However, since the "/DISPLAY" switch has been used, data will also be displayed on the terminal.