traceproto(8)

NAME

Traceproto - trace the route to a destination using a

specified protocol and port

SYNOPSIS

traceproto  [-cCTfAhvR]  [-p  protocol]  [-d dst_port] [-D
max_dst_port] [-s src_port] [-S max_src_port]  [-m  min_ttl]  [-M
max_ttl] [-w response_timeout] [-W send_delay] [-a account_level]
[-P    payload_size]    [-F    interface]    [-k    skips]    [-I
consecutive_trace_count]  [-H  packets_per_hop] [-i incr_pattern]
[-o output_style] [-t tcp_flags]

DESCRIPTION

traceproto is a replacement for the venerable traceroute.

It serves the same function, to determine the path that traffic

takes between the current host and a remote system. Like

traceroute it uses the IP time-to-live (TTL) field and watches

for the return ICMP packets from each successive gateway. In the

current Internet environment with ubiquitous firewalls and traf

fic filtering it is no longer enough to assume that if some traf

fic is getting to the remote machine that all of it is.

traceproto is an improvement to traceroute in this regard as it

lets you select the protocol and destination port to trace to.

OPTIONS

-p protocol

The protocol to use. The current options are tcp,

udp, and icmp with tcp as the default.

-d dst_port

The destination port to trace to. If -i is set to

'd' or 'D' then this is the minimum destination port. The de

fault value is 80 (HTTP).

-D max_dst_port

The maximum destination port. -i needs to be set

to 'd' or 'D' for this to have an effect.

-s src_port

The port to use as the local source port. If -i is

set to 's' or 'S' then it is the minimum source port. The de

fault is 10240.

-S max_src_port

The port to use as the maximum local source port.

-m min_ttl

Specifies the minimum time-to-live setting to use.

This sets how many hops away to start tracing. (For example, if

the first hop is a firewall that will not respond with an ICMP

TIME_EXCEEDED message but will allow packages through from hops

that are further away, you can set the min_ttl to 2 and thereby

avoid having to wait for the packets that timeout at the first

hop.) The default is 1.

-M Max_ttl

Specifies the maximum time-to-live setting. This

is the limit for how many hops away to trace. The default is 30.

(Very few locations on the Internet will be further away than

this.)

-w wait_timeout

The timeout in seconds to wait for a return packet.

The default is 5 seconds.

-W send_delay

The timeout before sending a new packet in mil

liseconds. The default is 100 milliseconds.

-a account_level

If accounting is enabled (ie. account_level set to

> 0) traceproto will print statistics after the run is complete

detailing minimum return time in milliseconds, average return

time in milliseconds, maximum return time in milliseconds, number

of packets returned successfully, and the number of packets that

generated no response. Possible account_levels:

0 No accounting printed.

1 Only the totals for all of the hops combined are

printed.

2 Stats are shown for each individual hop as well

as the total (this is the default).

Note that the -C flag will run traceproto in con

tinuous mode, restarting from scratch at the end of each run and

will generate stats after each run. Note also that CTRL-C will

abort the run, printing the stats at whatever level is specified

with -a and will abort traceproto. CTRL-Z will cause traceproto

to print the stats but does not abort the run.

-T Print timestamps with the output.

-P payload_size

This option specifies the payload size in bytes.

Currently the payload is filled with ' 's, though this may change

in the future. A word of caution, traceproto seems to have prob

lems with certain combinations of payload size and protocol, es

pecially udp with payloads of under 12 bytes. Hence the default

payload is 12 bytes. The cause of this has yet to be determined.

-k skips

Allows the user to specify a comma separated list

of hops to skip by number (ie. -k 5 will cause traceproto to

skip sending packets with a ttl of 5, going from 4 to 6). This

allows the user to skip a hop that is know to not respond, re

sponds slowly, or that should be avoided for some reason. The

skip list only allows comma separated hops at this point, not

ranges. Spaces in the list will also cause problems. Note that

-k 1 is functionally analogous to -m 2.

-c Trace continuously. This causes traceproto to re

set its counters and begin the trace from scratch once run has

been completed. Useful for finding intermittent problems and for

noting hops that use packet based routing (ie. the route used

for the packet may change from one packet to the next even during

the same connection). See also -C.

-C If -c is set, this prints the accumulated stats at

the end of each run. Meaningless unless -c is set.

-I consecutive_trace_count

Like -c and -C, -I does a continuous trace but un

like -c and -C, it only does consecutive_trace_count iterations.

-H packets_per_hop

This allows the user to specify the number of pack

ets sent out for each increment of the TTL. This must be a num

ber between 1 and 10. The default is 3 packets per hop.

-f Set the don't-fragment bit in the IP header. Can

be used in conjunction with the -P flag to determine at which hop

the MTR changes.

-F interface

Specify the network interface to use. Note that it

is quite possible to specify the wrong interface needed to reach

the target.

-A Do Autonomous System number lookups.

-i port_incr_pattern

This flag governs the source and destination incre

ment patterns. The pattern is a combination of one or two let

ters from the set [sSdDnN] where the meanings are:

s decrement the source port with each packet

S increment the source port with each packet (the

default for the source port)

d decrement the destination port with each packet

D increment the destination port with each packet

n use a static source port

N use a static destination port (the default for

the destination port)

Note that if contradictory options are given, the

last one seen is used (ie. -i Ss is the same as -i s). Note al

so that -i nN may cause traceproto problems in determining which

response is for which packet.

-o output_style

The default output style is designed to be reason

able legible and space conscious. There are other output styles

available:

s the standard output (default). Prints the hop

number, the response protocol and type, the response machine name

if available, and the response IP address on the first line. The

next line is a list of the response times as they arrive.

g graphical output. The first line is the same as

the standard output, but the response time is printed one line

per response with a variable length ascii bar indicating the rel

ative response time.

c classic output. An output style (almost) identi

cal to the original traceroute.

n no individual hop output. Stats will still print

at whatever level is set. (Note that -o n -a 0 is possible, but

silly.)

m minimal output. Prints the hop number and a sin

gle character indicating the response type. Response times are

not noted.

p scriptable output. Prints the hop number, ip,

return packet type, and time for each packet sent.

C (n)curses output.

-t flags

Sets the tcp flags to use.

S SYN (default)

A ACK

R RST

U URG

P PUSH

F FIN

E ECE

C CWR

Use -t SEC to test connections with Explicit Con

gestion Notification enabled.

-R Reverse trace. Starts at the max TTL and decre

ments the TTL for each hop.

-h A brief usage message summarizing the options.

-v Version info.

USAGE

The easiest way to invoke traceproto is simply to give it

a target machine by name or IP address. The default behavior is

to trace using tcp and destination port 80. Given the current

state of the Internet this is the most likely to succeed though

it certainly won't be appropriate for every situation.

traceproto assumes that the user has some understanding of

the protocols involved; users who are unfamiliar with them may

find that traceproto's options or output are hard to decipher.

Information on how the protocols function is widely available on

the Internet and is recommended reading for the traceproto user.

There are a fair number of options for traceproto and

there will probably be even more in the future. Many are contra

dictory or illogical in combination, but unless a combination is

actually problematic, traceproto will usually allow it to be

specified. Combinations that are illogical are generally ignored

in a sensible manner; contradictions are usually resolved using

the option furthest to the right on the command line.

ENVIRONMENT

There are a number of environmental variables that can be

set to effect traceproto's behavior. These variables override

the default behavior; in turn they are themselves are overridden

by their equivalent command line flags.

TP_TIMESTAMP_STYLE

Sets the format for the time-stamp printed by the

-T flag. Note that there is no equivalent command line flag for

this variable.

std The std day/alpha-month/year:hour:minute:second

format.

us US Style date of numeric

month/day/year:hour:minute:second format.

descend Descending time format:

year/month/day:hour:minute:second.

epoch Number of seconds since the unix epoch.

TP_DEFAULT_IF

Sets the default interface. Is overridden by the

-F flag. Setting this variable causes traceproto to print a mes

sage about which interface it is using, since traceproto will at

tempt to send the traces out that interface whether or not it is

the correct interface needed to reach the specified destination.

TP_OUTPUT_STYLE

Sets the default output style for traceproto. Is

overridden using the -o flag.

std The standard output style. The same as '-o s'.

graphical The graphical output style. The same as

'-o g'.

classic The classic output style. The same as '-o

c'.

none No output. The same as '-o n'.

minimal Minimal output. The same as '-o m'.

scriptable Output suitable for use in scripting.

The same as '-o p'.

curses (N)Curses output. The same as '-o C'.

RA_SERVER

and

TP_RA_SERVER

Sets the default server to to Autonomous System

number lookups. Mimics the NANOG traceroute behavior.

SEE ALSO

traceroute(8), hping2(8), tcptraceroute(1), mtr(8),

lft(8).

BUGS

None known at this time. Please report any bugs found.

traceproto March 12 2005

TRACEPROTO(8)