<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">
<chapter>
<header>
<copyright>
<year>2002</year><year>2016</year>
<holder>Ericsson AB. All Rights Reserved.</holder>
</copyright>
<legalnotice>
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
</legalnotice>
<title>Trace Tool Builder</title>
<prepared></prepared>
<docno></docno>
<date></date>
<rev></rev>
<file>ttb_ug.xml</file>
</header>
<section>
<title>Introduction</title>
<p>Trace Tool Builder is a base for building trace tools for
single node or distributed Erlang systems. It requires the
Runtime_Tools application to be available on the traced
node.
</p>
<p>The following are the main features of Trace Tool Builder:</p>
<list type="bulleted">
<item>Start tracing to file ports on many nodes with one
function call.</item>
<item>Write more information to a trace information file,
which is read during formatting.</item>
<item>Restore previous configuration by maintaining a
history buffer and handling configuration files.</item>
<item>Provide some simple support for sequential tracing.</item>
<item>Format binary trace logs and merge logs from
multiple nodes.</item>
</list>
<p>The intention of Trace Tool Builder is to serve
as a base for tailor-made trace tools, but it can also be used directly
from the Erlang shell (it can mimic <c>dbg</c> behaviour while
still providing useful additions, such as match specification shortcuts).
Trace Tool Builder only allows the use of file port tracer, so to use
other types of trace clients it is better to use <c>dbg</c> directly.</p>
</section>
<section>
<title>Getting Started</title>
<p>Module <c>ttb</c> is the interface to all functions in
Trace Tool Builder.</p>
<p>To get started, the least you need to do is to
start a tracer with
<seealso marker="ttb#tracer/0"><c>ttb:tracer/0,1,2</c></seealso>,
and set the required
trace flags on the processes you want to trace with
<seealso marker="ttb#p/2"><c>ttb:p/2</c></seealso>.</p>
<p>When the tracing is completed, stop the tracer with
<seealso marker="ttb#stop/0"><c>ttb:stop/0,1</c></seealso>
and format the trace log with
<seealso marker="ttb#format/1"><c>ttb:format/1,2</c></seealso>
(if there is anything to format).
</p>
<p><em>Useful functions:</em></p>
<taglist>
<tag><c>ttb:tracer/0,1,2</c></tag>
<item><p>Opens a trace port on each node to be traced. By default,
trace messages are written to binary files on remote nodes (the
binary trace log).</p></item>
<tag><c>ttb:p/2</c></tag>
<item><p>Specifies the processes to be traced. Trace flags specified
in this call specify what to trace on each process. This function can be
called many times if you like different trace flags to be set on different
processes.</p></item>
<tag><c>ttb:tp/2,3,4</c> or <c>ttb:tpl/2,3,4</c></tag>
<item><p>If you want to trace function calls (that is, if you have
trace flag <c>call</c> set on any process), you must
also set trace patterns on the required function(s) with
<seealso marker="ttb#/0"><c>ttb:tp/2,3,4</c></seealso> or
<seealso marker="ttb#/0"><c>ttb:tpl/2,3,4</c></seealso>.
A function is only traced
if it has a trace pattern. The trace pattern specifies how to trace the
function by using match specifications. Match specifications are
described in the
<seealso marker="erts:users_guide">ERTS User's Guide</seealso>.</p></item>
<tag><c>ttb:stop/0,1</c></tag>
<item><p>Stops tracing on all nodes, deletes all trace patterns, and
flushes the trace port buffer.</p></item>
<tag><c>ttb:format/1/2</c></tag>
<item><p>Translates the binary trace logs into something readable.
By default, <c>ttb</c> presents each trace message as a line of text,
but you can also write your own handler to make more complex interpretations
of the trace information. A trace log can also be presented graphically
with application Event Tracer (ET).</p>
<p>If option <c>format</c> is specified to <c>ttb:stop/1</c>, the formatting
is automatically done when stopping <c>ttb</c>.</p></item>
</taglist>
<section>
<title>Tracing Local Node from Erlang Shell</title>
<p>The following small module is used in the subsequent example:</p>
<code type="none">
-module(m).
-export([f/0]).
f() ->
receive
From when is_pid(From) ->
Now = erlang:now(),
From ! {self(),Now}
end. </code>
<p>The following example shows the basic use of <c>ttb</c> from
the Erlang shell. Default options are used both for starting the
tracer and for formatting (the custom fetch directory is however provided).
This gives a trace log named <c>Node-ttb</c> in the newly created
directory, where <c>Node</c> is the node name. The
default handler prints the formatted trace messages in the
shell:</p>
<pre>
(tiger@durin)47> %% First I spawn a process running my test function
(tiger@durin)47> <input>Pid = spawn(m,f,[]).</input>
<0.125.0>
(tiger@durin)48>
(tiger@durin)48> %% Then I start a tracer...
(tiger@durin)48> <input>ttb:tracer().</input>
{ok,[tiger@durin]}
(tiger@durin)49>
(tiger@durin)49> %% and activate the new process for tracing
(tiger@durin)49> %% function calls and sent messages.
(tiger@durin)49> <input>ttb:p(Pid,[call,send]).</input>
{ok,[{<0.125.0>,[{matched,tiger@durin,1}]}]}
(tiger@durin)50>
(tiger@durin)50> %% Here I set a trace pattern on erlang:now/0
(tiger@durin)50> %% The trace pattern is a simple match spec
(tiger@durin)50> %% indicating that the return value should be
(tiger@durin)50> %% traced. Refer to the reference_manual for
(tiger@durin)50> %% the full list of match spec shortcuts
(tiger@durin)50> %% available.
(tiger@durin)51> <input>ttb:tp(erlang,now,return).</input>
{ok,[{matched,tiger@durin,1},{saved,1}]}
(tiger@durin)52>
(tiger@durin)52> %% I run my test (i.e. send a message to
(tiger@durin)52> %% my new process)
(tiger@durin)52> <input>Pid ! self().</input>
<0.72.0>
(tiger@durin)53>
(tiger@durin)53> %% And then I have to stop ttb in order to flush
(tiger@durin)53> %% the trace port buffer
(tiger@durin)53> <input>ttb:stop([return, {fetch_dir, "fetch"}]).</input>
{stopped, "fetch"}
(tiger@durin)54>
(tiger@durin)54> %% Finally I format my trace log
(tiger@durin)54> <input>ttb:format("fetch").</input>
({<0.125.0>,{m,f,0},tiger@durin}) call erlang:now()
({<0.125.0>,{m,f,0},tiger@durin}) returned from erlang:now/0 ->
{1031,133451,667611}
({<0.125.0>,{m,f,0},tiger@durin}) <0.72.0> !
{<0.125.0>,{1031,133451,667611}}
ok</pre>
</section>
<section>
<title>Build Your Own Tool</title>
<p>The following example shows a simple tool for "debug tracing",
that is, tracing of function calls with return values:</p>
<code type="none"><![CDATA[
-module(mydebug).
-export([start/0,trc/1,stop/0,format/1]).
-export([print/4]).
%% Include ms_transform.hrl so that I can use dbg:fun2ms/2 to
%% generate match specifications.
-include_lib("stdlib/include/ms_transform.hrl").
%%% -------------Tool API-------------
%%% ----------------------------------
%%% Star the "mydebug" tool
start() ->
%% The options specify that the binary log shall be named
%% <Node>-debug_log and that the print/4 function in this
%% module shall be used as format handler
ttb:tracer(all,[{file,"debug_log"},{handler,{{?MODULE,print},0}}]),
%% All processes (existing and new) shall trace function calls
%% We want trace messages to be sorted upon format, which requires
%% timestamp flag. The flag is however enabled by default in ttb.
ttb:p(all,call).
%%% Set trace pattern on function(s)
trc(M) when is_atom(M) ->
trc({M,'_','_'});
trc({M,F}) when is_atom(M), is_atom(F) ->
trc({M,F,'_'});
trc({M,F,_A}=MFA) when is_atom(M), is_atom(F) ->
%% This match spec shortcut specifies that return values shall
%% be traced.
MatchSpec = dbg:fun2ms(fun(_) -> return_trace() end),
ttb:tpl(MFA,MatchSpec).
%%% Format a binary trace log
format(Dir) ->
ttb:format(Dir).
%%% Stop the "mydebug" tool
stop() ->
ttb:stop(return).
%%% --------Internal functions--------
%%% ----------------------------------
%%% Format handler
print(_Out,end_of_trace,_TI,N) ->
N;
print(Out,Trace,_TI,N) ->
do_print(Out,Trace,N),
N+1.
do_print(Out,{trace_ts,P,call,{M,F,A},Ts},N) ->
io:format(Out,
"~w: ~w, ~w:~n"
"Call : ~w:~w/~w~n"
"Arguments :~p~n~n",
[N,Ts,P,M,F,length(A),A]);
do_print(Out,{trace_ts,P,return_from,{M,F,A},R,Ts},N) ->
io:format(Out,
"~w: ~w, ~w:~n"
"Return from : ~w:~w/~w~n"
"Return value :~p~n~n",
[N,Ts,P,M,F,A,R]). ]]></code>
<p>To distinguish trace logs produced with this tool from other
logs, option <c>file</c> is used in
<seealso marker="ttb#tracer/2"><c>tracer/2</c></seealso>. The
logs are therefore fetched to a directory named
<c>ttb_upload_debug_log-YYYYMMDD-HHMMSS</c>
</p>
<p>By using option <c>handler</c> when starting the tracer,
the information about how to format the file is stored in the
trace information file (<c>.ti</c>). This is not necessary, as
it can be specified when formatting instead. However, It can
be useful if you, for example, want to format trace logs automatically
using option <c>format</c> in <c>ttb:stop/1</c>. Also, you do not need
any knowledge of the content of a binary log to format it the way it
is intended. If option <c>handler</c> is specified both when starting
the tracer and when formatting, the one specified when formatting is used.
</p>
<p>Trace flag <c>call</c> is set on all processes. This
means that any function activated with command <c>trc/1</c>
is traced on all existing and new processes.
</p>
</section>
</section>
<section>
<title>Running Trace Tool Builder against Remote Node</title>
<p>The Observer application might not always be available on the
node to be traced (in the following called the "traced
node"). However, Trace Tool Builder can still be run from
another node (in the following called the "trace control node") as
long as the following is fulfilled:
</p>
<list type="bulleted">
<item>The Observer application is available on the trace control node.</item>
<item>The Runtime_Tools application is available on both the
trace control node and the traced node.</item>
</list>
<p>If Trace Tool Builder is to be used against a remote node,
it is highly recommended to start the trace control node as
<em>hidden</em>. This way it can connect to the traced node
without being "seen" by it, that is, if the <c>nodes()</c>
BIF is called on the traced node, the trace control node does not
show. To start a hidden node, add option <c>-hidden</c> to the
<c>erl</c> command, for example:</p>
<pre>
% <input>erl -sname trace_control -hidden</input></pre>
<section>
<title>Diskless Node</title>
<p>If the traced node is diskless, <c>ttb</c> must be started from
a trace control node with disk access, and option <c>file</c>
must be specified to function <c>tracer/2</c> with value
<c>{local, File}</c>, for example:</p>
<pre>
(trace_control@durin)1> <input>ttb:tracer(mynode@diskless,
{file,{local,{wrap,"mytrace"}}}).</input>
{ok,[mynode@diskless]}</pre>
</section>
</section>
<section>
<title>More Tracing Options</title>
<p>When setting up a trace, the following features can also be activated:</p>
<list type="bulleted">
<item>Time-constrained tracing</item>
<item>Overload protection</item>
<item>Autoresume</item>
<item><c>dbg</c> mode</item>
</list>
<section>
<title>Time-Constrained Tracing</title>
<p>It can sometimes be helpful to enable trace for a
specified period of time (for example, to monitor a system for 24 hours
or half a second). This can be done with option
<c>{timer, TimerSpec}</c>. If <c>TimerSpec</c> has the
form of <c>MSec</c>, the trace is stopped after <c>MSec</c>
milliseconds using
<seealso marker="ttb#stop/0"><c>ttb:stop/0</c></seealso>. If more
options are provided (<c>TimerSpec = {MSec, Opts}</c>),
<seealso marker="ttb#stop/1"><c>ttb:stop/1</c></seealso>
is called instead with <c>Opts</c> as argument.</p>
<p>The timer is started with
<seealso marker="ttb#p/2"><c>ttb:p/2</c></seealso>, so any trace patterns
must be set up in advance.
<seealso marker="ttb#start_trace/4"><c>ttb:start_trace/4</c></seealso>
always sets up all patterns before invoking <c>ttb:p/2</c>.</p>
<p>The following example shows how to set up a trace that is
automatically stopped and formatted after 5 seconds:
</p><pre>
(tiger@durin)1> <input>ttb:start_trace([node()],
[{erlang, now,[]}],
{all, call},
[{timer, {5000, format}}]).</input></pre>
<note><p>Because of network and processing delays, the period
of tracing is approximate.</p></note>
</section>
<section>
<title>Overload Protection</title>
<p>When tracing live systems, always take special care to not
overload a node with too heavy tracing. <c>ttb</c> provides
option <c>overload</c> to address this problem.</p>
<p><c>{overload, MSec, Module, Function}</c> instructs the <c>ttb</c> back end
(a part of the <seealso marker="runtime_tools:index">Runtime_Tools</seealso>
application) to perform overload check every <c>MSec</c> millisecond.
If the check (named <c>Module:Function(check)</c>) returns
<c>true</c>, tracing is disabled on the selected node.</p>
<p>Overload protection activated on one node does not
affect other nodes, where the tracing continues as normal.
<c>ttb:stop/0,1</c> fetches data from all clients, including everything
collected before the activation of overload protection.</p>
<note><p>
It is not allowed to change trace details
(with <c>ttb:p</c> and <c>ttb:tp/tpl...</c>) once overload
protection is activated in one of the traced nodes. This is to
avoid trace setup being inconsistent between nodes.</p></note>
<p><c>Module:Function</c> provided with option <c>overload</c> must
handle three calls: <c>init</c>, <c>check</c>, and <c>stop</c>. <c>init</c>
and <c>stop</c> allow some setup and teardown required by
the check. An overload check module can look as follows:
</p><code type="none">
-module(overload).
-export([check/1]).
check(init) ->
Pid = sophisticated_module:start(),
put(pid, Pid);
check(check) ->
get(pid) ! is_overloaded,
receive
Reply ->
Reply
after 5000 ->
true
end;
check(stop) ->
get(pid) ! stop.</code>
<note><p>
<c>check</c> is always called by the same process, so <c>put</c> and
<c>get</c> are possible.</p></note>
</section>
<section>
<title>Autoresume</title>
<p>A node can crash (probably a buggy one, hence traced).
Use <c>resume</c> to resume tracing on the node automatically
when it gets back. The failing node then tries to reconnect
to trace control node when <c>Runtime_Tools</c> is started.
This implies that <c>Runtime_Tools</c> must be included in
the startup chain of other nodes (if not, you can still
resume tracing by starting <c>Runtime_Tools</c> manually,
that is, by an RPC call).</p>
<p>To not lose the data that the failing node stored
up to the point of crash, the control node tries to fetch
it before restarting trace. This must occur within the allowed
time frame, otherwise it is aborted (default is 10 seconds, but it
can be changed with <c>{resume, MSec}</c>). The data fetched
this way is then merged with all other traces.</p>
<p>The autostart feature requires more data to be stored on
traced nodes. By default, the data is stored automatically
to the file named "ttb_autostart.bin" in the currect working directory
(cwd) of the traced node.
Users can change this behaviour (that is, on diskless
nodes) by specifying their own module to handle autostart data
storage and retrieval (<c>ttb_autostart_module</c>
environment variable of <c>runtime_tools</c>). For information
about the API, see module
<seealso marker="ttb"><c>ttb</c></seealso>.
The following example shows the default handler:</p>
<code>
-module(ttb_autostart).
-export([read_config/0,
write_config/1,
delete_config/0]).
-define(AUTOSTART_FILENAME, "ttb_autostart.bin").
delete_config() ->
file:delete(?AUTOSTART_FILENAME).
read_config() ->
case file:read_file(?AUTOSTART_FILENAME) of
{ok, Data} -> {ok, binary_to_term(Data)};
Error -> Error
end.
write_config(Data) ->
file:write_file(?AUTOSTART_FILENAME, term_to_binary(Data)).</code>
<note><p>Remember that file trace ports buffer the data
by default. If the node crashes, trace messages are not
flushed to the binary log. If the risk of failure is
high, it can be a good idea to flush the buffers every
now and then automatically. Passing <c>{flush, MSec}</c>
as an option of <c>ttb:tracer/2</c> flushes all buffers
every <c>MSec</c> millisecond.</p></note>
</section>
<section>
<title>dbg Mode</title>
<p>Option <c>{shell, ShellType}</c> allows making <c>ttb</c>
operation similar to
<seealso marker="runtime_tools:dbg"><c>dbg</c></seealso>.
Using <c>{shell, true}</c>
displays all trace messages in the shell before storing them.
<c>{shell, only}</c> additionally disables message storage
(making the tool to behave exactly like <c>dbg</c>). This is
allowed only with IP trace ports (<c>{trace, {local, File}}</c>).
</p>
<p>Command <c>ttb:tracer(dbg)</c> is a shortcut for the pure
<c>dbg</c> mode (<c>{shell, only}</c>).</p>
</section>
</section>
<section>
<marker id="trace_info"></marker>
<title>Trace Information and File .ti</title>
<p>In addition to the trace log file(s), a file with extension
<c>.ti</c> is created when Trace Tool Builder is started. This
is the trace information file. It is a binary file, which
contains the process information, trace flags used, the name of
the node to which it belongs, and all information written with
function
<seealso marker="ttb#write_trace_info/2"><c>ttb:write_trace_info/2</c></seealso>.
<c>.ti</c> files are always fetched with other logs when the trace is stopped.
</p>
<p>Except for the process information, everything in the trace
information file is passed on to the handler function when
formatting. Parameter <c>TI</c> is a list of
<c>{Key,ValueList}</c> tuples. The keys <c>flags</c>,
<c>handler</c>, <c>file</c>, and <c>node</c> are used for
information written directly by <c>ttb</c>.
</p>
<p>Information to the trace information file by
can be added by calling
<seealso marker="ttb#write_trace_info/2"><c>ttb:write_trace_info/2</c></seealso>.
Notice that <c>ValueList</c>
always is a list, and if you call <c>write_trace_info/2</c>
many times with the same <c>Key</c>, the <c>ValueList</c> is
extended with a new value each time.
</p>
<p><em>Example:</em></p>
<p><c>ttb:write_trace_info(mykey,1)</c> gives the entry
<c>{mykey,[1]}</c> in <c>TI</c>. Another call,
<c>ttb:write_trace_info(mykey,2)</c>, changes this entry to
<c>{mykey,[1,2]}</c>.
</p>
</section>
<section>
<title>Wrap Logs</title>
<p>If you want to limit the size of the trace logs, you can use
wrap logs. This works almost like a circular buffer. You can
specify the maximum number of binary logs and the maximum size of
each log. <c>ttb</c> then creates a new binary log each time a log
reaches the maximum size. When the maximum number of logs are
reached, the oldest log is deleted before a new one is created.
</p>
<note><p>The overall size of data generated by <c>ttb</c> can be greater
than the wrap specification suggests. If a traced node restarts
and autoresume is enabled, the old wrap log is always stored and
a new one is created.
</p></note>
<p>Wrap logs can be formatted one by one or all at once. See
<seealso marker="#format">Formatting</seealso>.
</p>
</section>
<section>
<marker id="format"></marker>
<title>Formatting</title>
<p>Formatting can be done automatically when stopping <c>ttb</c>
(see section
<seealso marker="#fetch_format">Automatically Collect and Format Logs from All Nodes</seealso>), or explicitly by calling function
<c>ttb:format/1,2</c>.
</p>
<p>Formatting means to read a binary log and present it in a
readable format. You can use the default format handler in
<c>ttb</c> to present each trace message as a line of text, or
write your own handler to make more complex interpretations of the
trace information. You can also use application ET to
present the trace log graphically (see section
<seealso marker="#et_viewer">Presenting Trace Logs with Event Tracer</seealso>).
</p>
<p>The first argument to <c>ttb:format/1,2</c> specifies which
binary log(s) to format. This is usually the name of a directory
that <c>ttb</c> created during log fetch. Unless option
<c>disable_sort</c> is provided, the logs from different files
are always sorted according to time-stamp in traces.
</p>
<p>The second argument to <c>ttb:format/2</c> is a list of
options as follows:
</p>
<taglist>
<tag><c>out</c></tag>
<item><p>Specifies the destination to write the formatted text.
Default destination is <c>standard_io</c>, but a filename can
also be specified.</p></item>
<tag><c>handler</c></tag>
<item><p>Specifies the format handler to use. If this option is
not specified, option <c>handler</c> that is specified when starting
the tracer is used. If option <c>handler</c> is not specified
when starting the tracer either, a default handler is used, which
prints each trace message as a text line.</p></item>
<tag><c>disable_sort</c></tag>
<item><p>Indicates that the logs are not to be merged according to
time-stamp, but processed one file after another (this can be
a bit faster).</p></item>
</taglist>
<p>A format handler is a fun taking four arguments. This fun is
called for each trace message in the binary log(s). A simple
example that only prints each trace message can be as follows:</p>
<code type="none">
fun(Fd, Trace, _TraceInfo, State) ->
io:format(Fd, "Trace: ~p~n", [Trace]),
State
end. </code>
<p>Here, <c>Fd</c> is the file descriptor for the destination file, or
the atom <c>standard_io</c>. <c>_TraceInfo</c> contains information
from the trace information file (see section
<seealso marker="#trace_info">Trace Information and File .ti</seealso>). <c>State</c> is a state variable for the format
handler fun. The initial value of variable <c>State</c> is
specified with the handler option, for example:</p>
<code type="none">
ttb:format("tiger@durin-ttb", [{handler, {{Mod,Fun}, initial_state}}])
^^^^^^^^^^^^^ </code>
<p>Another format handler can be used to calculate the time spent by
the garbage collector:</p>
<code type="none">
fun(_Fd,{trace_ts,P,gc_start,_Info,StartTs},_TraceInfo,State) ->
[{P,StartTs}|State];
(Fd,{trace_ts,P,gc_end,_Info,EndTs},_TraceInfo,State) ->
{value,{P,StartTs}} = lists:keysearch(P,1,State),
Time = diff(StartTs,EndTs),
io:format("GC in process ~w: ~w milliseconds~n", [P,Time]),
State -- [{P,StartTs}]
end </code>
<p>A more refined version of this format handler is function
<c>handle_gc/4</c> in module <c>multitrace.erl</c>
included in directory <c>src</c> of the Observer application.
</p>
<p>The trace message is passed as the second argument (<c>Trace</c>).
The possible values of <c>Trace</c> are the following:</p>
<list type="bulleted">
<item>All trace messages described in
<seealso marker="erts:erlang#trace/3"><c>erlang:trace/3</c></seealso>
</item>
<item><c>{drop, N}</c> if IP tracer is used (see
<seealso marker="runtime_tools:dbg#trace_port/2"><c>dbg:trace_port/2</c></seealso>)
</item>
<item><c>end_of_trace</c> received once when all trace messages are
processed</item>
</list>
<p>By giving the format handler
<seealso marker="ttb#get_et_handler/0"><c>ttb:get_et_handler()</c></seealso>,
you can have the trace
log presented graphically with <c>et_viewer</c> in the ET
application (see section
<seealso marker="#et_viewer">Presenting Trace Logs with Event Tracer</seealso>).
</p>
<p>You can always decide not to format the whole trace data contained
in the fetch directory, but analyze single files instead. To do so,
a single file (or list of files) must be passed as the first argument
to <c>format/1,2</c>.</p>
<p>Wrap logs can be formatted one by one or all at once. To
format one of the wrap logs in a set, specify the exact file name.
To format the whole set of wrap logs, specify the name with <c>*</c>
instead of the wrap count.
</p>
<p><em>Example:</em></p>
<p>Start tracing:</p>
<pre>
(tiger@durin)1> <input>ttb:tracer(node(),{file,{wrap,"trace"}}).</input>
{ok,[tiger@durin]}
(tiger@durin)2> <input>ttb:p(...)</input>
...</pre>
<p>This gives a set of binary logs, for example:</p>
<code type="none">
[email protected]
[email protected]
[email protected]
... </code>
<p>Format the whole set of logs:</p>
<pre>
1> <input>ttb:format("tiger@durin-trace.*.wrp").</input>
....
ok
2> </pre>
<p>Format only the first log:</p>
<pre>
1> <input>ttb:format("[email protected]").</input>
....
ok
2> </pre>
<p>To merge all wrap logs from two nodes:</p>
<pre>
1> <input>ttb:format(["tiger@durin-trace.*.wrp","lion@durin-trace.*.wrp"]).</input>
....
ok
2> </pre>
<section>
<marker id="et_viewer"></marker>
<title>Presenting Trace Logs with Event Tracer</title>
<p>For detailed information about the Event Tracer, see the
<seealso marker="et:users_guide">ET</seealso> application.
</p>
<p>By giving the format handler
<seealso marker="ttb#get_et_handler/0"><c>ttb:get_et_handler()</c></seealso>,
you can have the trace log presented graphically with
<c>et_viewer</c> in the ET application.
<c>ttb</c> provides filters that can be selected from the
menu <em>Filter</em> in the <c>et_viewer</c> window. The filters
are names according to the type of actors they present
(that is, what each vertical line in the sequence diagram represents).
Interaction between actors is shown as red arrows between two
vertical lines, and activities within an actor are shown as
blue text to the right of the actors line.
</p>
<p>The <c>processes</c> filter is the only filter showing all
trace messages from a trace log. Each vertical line in
the sequence diagram represents a process. Erlang messages,
spawn, and link/unlink are typical interactions between
processes. Function calls, scheduling, and garbage collection,
are typical activities within a process. <c>processes</c> is
the default filter.
</p>
<p>The remaining filters only show function calls and
function returns. All other trace message are discarded. To get
the most out of these filters, <c>et_viewer</c> must know
the caller of each function and the time of return. This can be
obtained using both the <c>call</c> and <c>return_to</c>
flags when tracing. Notice that flag <c>return_to</c> only
works with local call trace, that is, when trace patterns are set
with <c>ttb:tpl</c>.
</p>
<p>The same result can be obtained by using the flag <c>call</c>
only and setting a match specification on local or
global function calls as follows:</p>
<pre>
1> <input>dbg:fun2ms(fun(_) -> return_trace(),message(caller()) end).</input>
[{'_',[],[{return_trace},{message,{caller}}]}]</pre>
<p>This must however be done with care, as function
<c>{return_trace}</c> in the match specification
destroys tail recursiveness.
</p>
<p>The <c>modules</c> filter shows each module as a vertical
line in the sequence diagram. External function calls/returns
are shown as interactions between modules, and internal function
calls/returns are shown as activities within a module.
</p>
<p>The <c>functions</c> filter shows each function as a vertical
line in the sequence diagram. A function calling itself is shown
as an activity within a function, and all other function calls
are shown as interactions between functions.
</p>
<p>The <c>mods_and_procs</c> and <c>funcs_and_procs</c> filters
are equivalent to the <c>modules</c> and <c>functions</c>
filters respectively, except that each module or function can
have many vertical lines, one for each process it resides on.
</p>
<p>In the following example, modules <c>foo</c> and <c>bar</c> are used:</p>
<code type="none">
-module(foo).
-export([start/0,go/0]).
start() ->
spawn(?MODULE, go, []).
go() ->
receive
stop ->
ok;
go ->
bar:f1(),
go()
end.</code>
<code type="none">
-module(bar).
-export([f1/0,f3/0]).
f1() ->
f2(),
ok.
f2() ->
spawn(?MODULE,f3,[]).
f3() ->
ok.</code>
<p>Setting up the trace:</p>
<pre>
(tiger@durin)1> %%First we retrieve the Pid to limit traced processes set
(tiger@durin)1> <input>Pid = foo:start().</input>
(tiger@durin)2> %%Now we set up tracing
(tiger@durin)2> <input>ttb:tracer().</input>
(tiger@durin)3> <input>ttb:p(Pid, [call, return_to, procs, set_on_spawn]).</input>
(tiger@durin)4> <input>ttb:tpl(bar, []).</input>
(tiger@durin)5> %%Invoke our test function and see output with et viewer
(tiger@durin)5> <input>Pid ! go.</input>
(tiger@durin)6> <input>ttb:stop({format, {handler, ttb:get_et_handler()}}).</input></pre>
<p>This renders a result similar to the following:
</p>
<image file="et_processes.gif">
<icaption>Filter: "processes"</icaption>
</image>
<p></p>
<image file="et_modsprocs.gif">
<icaption>Filter: "mods_and_procs"</icaption>
</image>
<p>Notice that function
<seealso marker="ttb#start_trace/4"><c>ttb:start_trace/4</c></seealso>
can be used as help as follows:</p>
<pre>
(tiger@durin)1> <input>Pid = foo:start().</input>
(tiger@durin)2> <input>ttb:start_trace([node()],
[{bar,[]}],
{Pid, [call, return_to, procs, set_on_spawn]}
{handler, ttb:get_et_handler()}).</input>
(tiger@durin)3> <input>Pid ! go.</input>
(tiger@durin)4> <input>ttb:stop(format).</input></pre>
</section>
</section>
<section>
<marker id="fetch_format"></marker>
<title>Automatically Collect and Format Logs from All Nodes</title>
<p>By default,
<seealso marker="ttb#stop/1"><c>ttb:stop/1</c></seealso> fetches trace logs
and trace information files from all nodes. The logs are stored in a
new directory named <c>ttb_upload-Filename-Timestamp</c> under the
working directory of the trace control node. Fetching can be disabled
by providing option <c>nofetch</c> to <c>ttb:stop/1</c>. The user can
specify a fetch directory by passing option <c>{fetch_dir, Dir}</c>.
</p>
<p>If option <c>format</c> is specified to <c>ttb:stop/1</c>, the
trace logs are automatically formatted after tracing is
stopped.
</p>
</section>
<section>
<title>History and Configuration Files</title>
<p>For the tracing functionality,
<seealso marker="runtime_tools:dbg"><c>dbg</c></seealso>
can be used instead
of <c>ttb</c> for setting trace flags on processes and trace
patterns for call trace, that is, the functions
<c>p</c>, <c>tp</c>, <c>tpl</c>, <c>ctp</c>, <c>ctpl</c>, and <c>ctpg</c>. Only the
following two things are added by <c>ttb</c> for these functions:</p>
<list type="bulleted">
<item>All calls are stored in the history buffer and can be
recalled and stored in a configuration file. This makes it
easy to set up the same trace environment, for example, if you
want to compare two test runs. It also reduces the amount of
typing when using <c>ttb</c> from the Erlang shell.</item>
<item>Shortcuts are provided for the most common match
specifications (to not force you to use
<seealso marker="runtime_tools:dbg#fun2ms/1"><c>dbg:fun2ms</c></seealso>
continually).</item>
</list>
<p>Use
<seealso marker="ttb#list_history/0"><c>ttb:list_history/0</c></seealso>
to see the content of the history buffer and
<seealso marker="ttb#run_history/1"><c>ttb:run_history/1</c></seealso>
to re-execute one of the entries.
</p>
<p>The main purpose of the history buffer is the possibility to
create configuration files. Any function stored in the history
buffer can be written to a configuration file and used for
creating a specific configuration at any time with a single
function call.
</p>
<p>A configuration file is created or extended with
<seealso marker="ttb#write_config/2"><c>ttb:write_config/2,3</c></seealso>.
Configuration files are binary files
and can therefore only be read and written with functions provided
by <c>ttb</c>.
</p>
<p>The complete content of the history buffer can be written to a
configuration file by calling
<c>ttb:write_config(ConfigFile,all)</c>. Selected entries from
the history can be written by calling
<c>ttb:write_config(ConfigFile,NumList)</c>, where
<c>NumList</c> is a list of integers pointing out the history
entries to write. Moreover, the history buffer is always dumped
to <c>ttb_last_config</c> when <c>ttb:stop/0,1</c> is called.
</p>
<p>User-defined entries can also be written to a configuration file
by calling function
<c>ttb:write_config(ConfigFile,ConfigList)</c>, where
<c>ConfigList</c> is a list of <c>{Module,Function,Args}</c>.
</p>
<p>Any existing file <c>ConfigFile</c> is deleted and a new file
is created when <c>write_config/2</c> is called. Option
<c>append</c> can be used to add something at the end
of an existing configuration file, for example,
<c>ttb:write_config(ConfigFile,What,[append])</c>.
</p>
<p><em>Example:</em></p>
<p>See the content of the history buffer:</p>
<pre>
(tiger@durin)191> <input>ttb:tracer().</input>
{ok,[tiger@durin]}
(tiger@durin)192> <input>ttb:p(self(),[garbage_collection,call]).</input>
{ok,{[<0.1244.0>],[garbage_collection,call]}}
(tiger@durin)193> <input>ttb:tp(ets,new,2,[]).</input>
{ok,[{matched,1}]}
(tiger@durin)194> <input>ttb:list_history().</input>
[{1,{ttb,tracer,[tiger@durin,[]]}},
{2,{ttb,p,[<0.1244.0>,[garbage_collection,call]]}},
{3,{ttb,tp,[ets,new,2,[]]}}]</pre>
<p>Execute an entry from the history buffer:</p>
<pre>
(tiger@durin)195> <input>ttb:ctp(ets,new,2).</input>
{ok,[{matched,1}]}
(tiger@durin)196> <input>ttb:list_history().</input>
[{1,{ttb,tracer,[tiger@durin,[]]}},
{2,{ttb,p,[<0.1244.0>,[garbage_collection,call]]}},
{3,{ttb,tp,[ets,new,2,[]]}},
{4,{ttb,ctp,[ets,new,2]}}]
(tiger@durin)197> <input>ttb:run_history(3).</input>
ttb:tp(ets,new,2,[]) ->
{ok,[{matched,1}]}</pre>
<p>Write the content of the history buffer to a configuration
file:</p>
<pre>
(tiger@durin)198> <input>ttb:write_config("myconfig",all).</input>
ok
(tiger@durin)199> <input>ttb:list_config("myconfig").</input>
[{1,{ttb,tracer,[tiger@durin,[]]}},
{2,{ttb,p,[<0.1244.0>,[garbage_collection,call]]}},
{3,{ttb,tp,[ets,new,2,[]]}},
{4,{ttb,ctp,[ets,new,2]}},
{5,{ttb,tp,[ets,new,2,[]]}}]</pre>
<p>Extend an existing configuration:</p>
<pre>
(tiger@durin)200> <input>ttb:write_config("myconfig",[{ttb,tp,[ets,delete,1,[]]}],
[append]).</input>
ok
(tiger@durin)201> <input>ttb:list_config("myconfig").</input>
[{1,{ttb,tracer,[tiger@durin,[]]}},
{2,{ttb,p,[<0.1244.0>,[garbage_collection,call]]}},
{3,{ttb,tp,[ets,new,2,[]]}},
{4,{ttb,ctp,[ets,new,2]}},
{5,{ttb,tp,[ets,new,2,[]]}},
{6,{ttb,tp,[ets,delete,1,[]]}}]</pre>
<p>Go back to a previous configuration after stopping Trace Tool
Builder:</p>
<pre>
(tiger@durin)202> <input>ttb:stop().</input>
ok
(tiger@durin)203> <input>ttb:run_config("myconfig").</input>
ttb:tracer(tiger@durin,[]) ->
{ok,[tiger@durin]}
ttb:p(<0.1244.0>,[garbage_collection,call]) ->
{ok,{[<0.1244.0>],[garbage_collection,call]}}
ttb:tp(ets,new,2,[]) ->
{ok,[{matched,1}]}
ttb:ctp(ets,new,2) ->
{ok,[{matched,1}]}
ttb:tp(ets,new,2,[]) ->
{ok,[{matched,1}]}
ttb:tp(ets,delete,1,[]) ->
{ok,[{matched,1}]}
ok</pre>
<p>Write selected entries from the history buffer to a
configuration file:</p>
<pre>
(tiger@durin)204> <input>ttb:list_history().</input>
[{1,{ttb,tracer,[tiger@durin,[]]}},
{2,{ttb,p,[<0.1244.0>,[garbage_collection,call]]}},
{3,{ttb,tp,[ets,new,2,[]]}},
{4,{ttb,ctp,[ets,new,2]}},
{5,{ttb,tp,[ets,new,2,[]]}},
{6,{ttb,tp,[ets,delete,1,[]]}}]
(tiger@durin)205> <input>ttb:write_config("myconfig",[1,2,3,6]).</input>
ok
(tiger@durin)206> <input>ttb:list_config("myconfig").</input>
[{1,{ttb,tracer,[tiger@durin,[]]}},
{2,{ttb,p,[<0.1244.0>,[garbage_collection,call]]}},
{3,{ttb,tp,[ets,new,2,[]]}},
{4,{ttb,tp,[ets,delete,1,[]]}}]
(tiger@durin)207></pre>
</section>
<section>
<title>Sequential Tracing</title>
<p>To learn what sequential tracing is and how it can be used,
see the Reference Manual for
<seealso marker="kernel:seq_trace"><c>seq_trace</c></seealso>.
</p>
<p>The support for sequential tracing provided by Trace Tool
Builder includes the following:</p>
<list type="bulleted">
<item>Initiation of the system tracer. This is automatically
done when a trace port is started with
<seealso marker="ttb#tracer/0"><c>ttb:tracer/0,1,2</c></seealso>.</item>
<item>Creation of match specifications that activates
sequential tracing.</item>
</list>
<p>Starting sequential tracing requires that a tracer is
started with function <c>ttb:tracer/0,1,2</c>. Sequential
tracing can then be started in either of the following ways:
</p>
<list type="bulleted">
<item>Through a trigger function with a match specification
created with
<seealso marker="ttb#seq_trigger_ms/0"><c>ttb:seq_trigger_ms/0,1</c></seealso>.</item>
<item>Directly by using module
<seealso marker="kernel:seq_trace"><c>seq_trace</c></seealso>.</item>
</list>
<p><em>Example 1:</em></p>
<p>In the following example, function
<c>dbg:get_tracer/0</c> is used as trigger for sequential
tracing:</p>
<pre>
(tiger@durin)110> <input>ttb:tracer().</input>
{ok,[tiger@durin]}
(tiger@durin)111> <input>ttb:p(self(),call).</input>
{ok,{[<0.158.0>],[call]}}
(tiger@durin)112> <input>ttb:tp(dbg,get_tracer,0,ttb:seq_trigger_ms(send)).</input>
{ok,[{matched,1},{saved,1}]}
(tiger@durin)113> <input>dbg:get_tracer(), seq_trace:reset_trace().</input>
true
(tiger@durin)114> <input>ttb:stop(format).</input>
({<0.158.0>,{shell,evaluator,3},tiger@durin}) call dbg:get_tracer()
SeqTrace [0]: ({<0.158.0>,{shell,evaluator,3},tiger@durin})
{<0.237.0>,dbg,tiger@durin} ! {<0.158.0>,{get_tracer,tiger@durin}}
[Serial: {0,1}]
SeqTrace [0]: ({<0.237.0>,dbg,tiger@durin})
{<0.158.0>,{shell,evaluator,3},tiger@durin} ! {dbg,{ok,#Port<0.222>}}
[Serial: {1,2}]
ok
(tiger@durin)116></pre>
<p><em>Example 2:</em></p>
<p>Starting sequential tracing with a trigger is more
useful if the trigger function is not called directly from the
shell, but rather implicitly within a larger system. When
calling a function from the shell, it is simpler to start
sequential tracing directly, for example, as follows:</p>
<pre>
(tiger@durin)116> <input>ttb:tracer().</input>
{ok,[tiger@durin]}
(tiger@durin)117> <input>seq_trace:set_token(send,true), dbg:get_tracer(),
seq_trace:reset_trace().</input>
true
(tiger@durin)118> <input>ttb:stop(format).</input>
SeqTrace [0]: ({<0.158.0>,{shell,evaluator,3},tiger@durin})
{<0.246.0>,dbg,tiger@durin} ! {<0.158.0>,{get_tracer,tiger@durin}}
[Serial: {0,1}]
SeqTrace [0]: ({<0.246.0>,dbg,tiger@durin})
{<0.158.0>,{shell,evaluator,3},tiger@durin} ! {dbg,{ok,#Port<0.229>}}
[Serial: {1,2}]
ok
(tiger@durin)120></pre>
<p>In both previous examples, <c>seq_trace:reset_trace/0</c>
resets the trace token immediately after the traced function
to avoid many trace messages because of the printouts in
the Erlang shell.
</p>
<p>All functions in module <c>seq_trace</c>, except
<c>set_system_tracer/1</c>, can be used after the trace port
is started with <c>ttb:tracer/0,1,2</c>.
</p>
</section>
<section>
<title>Multipurpose Trace Tool</title>
<p>Module <c>multitrace</c> in
directory <c>src</c> of the Observer application provides a
small tool with three possible trace settings. The trace messages
are written to binary files, which can be formatted with
function <c>multitrace:format/1,2</c>:
</p>
<taglist>
<tag><c>multitrace:debug(What)</c></tag>
<item><p>Start calltrace on all processes and trace the specified
function(s). The format handler used is
<c>multitrace:handle_debug/4</c> that prints each call and
returns. <c>What</c> must be an item or a list of items to trace,
specified on the format <c>{Module,Function,Arity}</c>,
<c>{Module,Function}</c>, or only <c>Module</c>.</p></item>
<tag><c>multitrace:gc(Procs)</c></tag>
<item><p>Trace garbage collection on the specified process(es). The
format handler used is <c>multitrace:handle_gc/4</c> that
prints start, stop, and the time spent for each garbage collection.</p></item>
<tag><c>multitrace:schedule(Procs)</c></tag>
<item><p>Trace in-scheduling and out-scheduling on the specified process(es).
The format handler used is <c>multitrace:handle_schedule/4</c> that
prints each in-scheduling and out-scheduling with process, time-stamp, and
current function. It also prints the total time each traced
process was scheduled in.</p></item>
</taglist>
</section>
</chapter>
