6.0;probenet (probe_network), revision 6.0, 83/03/10
PROBENET (PROBE_NETWORK) -- Probe network and display error statistics.
usage: PROBENET [-A | -N <node-ids> | -T <topology_file>]
[{-L|-ERR|-Q|-MON|-I|-S <cnt>|-LEN
<size>|-R <secs>}]
FORMAT
PROBENET [options]
This command sends packets to a diagnostic socket in the target node and
receives replies indicating the status the request was received with. It
compiles statistics on the error rates to each node and reports them to
standard output.
OPTIONS
Default options are indicated by "(D)."
The following options specify the list of nodes to test.
-A (D) Probe all nodes responding to an LCNODE command. If the
network is completely corrupted so that messages cannot
make a complete pass, use one of the other two options to
specify precisely which nodes to test.
-N node_id ...
Probe the node(s) specified by the indicated hexadecimal
node ID(s). A good choice of nodes to test is a set evenly
spaced around the network.
-T pathname Probe the nodes listed in the topology file indicated. The
file must contain one hexadecimal node ID per line. Any
text following a space after the node ID is ignored.
Comment lines may be inserted if they are prefixed with a
' ' (space) or '{'.
The following options control the level of detail in the
statistics report printed.
-L Print long (detailed) error counts if there were any errors
(i.e., at least one XMIT ERRS or RCV ERRS).
-ERR Print header for each test, but only statistics for nodes
which returned errors (XMIT and/or RCV ERRS).
-Q Print only header and statistics for tests which returned
errors. Implies -ERR.
-MON Change output format to wide display, and print only tests
which have errors indicating that the whole network may be
failing. Normally, this is used in conjuction with the -N
node_id and -R [secs] options.
The following options modify the test which is run.
-I Test each node individually by sending all test packets to
each node one after the other in turn, and print out stats
after each node is tested. If this option is omitted,
PROBENET sends 1 packet to all nodes, then the next packet
to all nodes, until the number of packets specified by -S
has been reached.
-S n (D) Specify the total number of packets to be sent to each
node. The default number of packets is 10.
-LEN n (D) Specify the length of the data portion of the test packet,
in bytes. The data pattern is always FFFF FFFF 3333 3333
(repeated). The default length is 1024 bytes.
-R [n] Repeat PROBENET command every 'n' seconds. If 'n' is
omitted, the probe is repeated every 10 seconds. End the
probe with CTRL/Q.
EXAMPLES
1. $ probenet Probe the entire network. No errors detected.
83/03/31 9:02:50 Node 1FB: Sending 10 1024-byte packets to
92 nodes all at once.
Last ring hardware failure detected by node 3D4 on 83/03/30 at 20:02
XMIT XMIT XMIT RCV RCV
NODE NAME % REP %GOOD ACPT IGN ERRS REPS ERRS
---- ------------ ------ ------ ----- ----- ----- ----- -----
1FB unknown Self
83 BOGART 100.00 100.00 10 0 0 10 0
.
.
.
368 unknown 100.00 100.00 10 0 0 10 0
74 DEAN 100.00 100.00 10 0 0 10 0
2. $ probenet -q Probe quietly. No errors detected.
$
3. $ probenet -n 74 Probe node 74.
83/03/31 9:05:59 Node 1FB: Sending 10 1024-byte packets to 1 node.
Last ring hardware failure detected by node 3D4 on 83/03/30 at 20:02
XMIT XMIT XMIT RCV RCV
NODE NAME % REP %GOOD ACPT IGN ERRS REPS ERRS
---- ------------ ------ ------ ----- ----- ----- ----- -----
74 DEAN 100.00 100.00 10 0 0 10 0
REPORTED STATISTICS
The following statistics are printed for each node:
% REP - Percent of transmits for which replies were received.
This is the number of accepted transmits for which PROBENET
received replies. It should be 100%. Numbers < 100% indicate
that the remote node is having trouble delivering replies.
A very low number indicates that there may not be a sufficient
sample for the '% GOOD' statistic to be valid.
Specifically this is the ratio of Receive replies
to Accepted transmits.
% GOOD - Percent of replies which indicated the request was received
with a good or 'runt' status.
This number indicates how good the transmission channel is to
the specified node. It should be 100%. This number should
be < 100% for nodes with problems, or nodes downstream of a
'corrupting influence' on the network. Specifically this is
the ratios of Receive errors to Receive replies.
XMIT ACPT - Accepted transmits.
The number of transmits which the remote node should have
seen and sent replies to.
XMIT IGN - Ignored transmits.
The number of transmits which the remote node did not see.
Specifically this is the total of NACKs and WACKs.
XMIT ERRS - Transmit errors.
The number of accepted transmits which had some sort
of transmit error.
RCV REPS - Receive replies.
The number of replies received from the remote node.
RCV ERRS - Receive errors.
The number of replies which indicated that the remote
node received the packet with an error.