<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">
<chapter>
<header>
<copyright>
<year>2011</year><year>2017</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>Observer</title>
<prepared></prepared>
<docno></docno>
<date></date>
<rev></rev>
<file>observer_ug.xml</file>
</header>
<section>
<title>Introduction</title>
<p>Observer is a graphical tool for observing the characteristics of
Erlang systems. Observer displays system information, application
supervisor trees, process information, ETS tables, Mnesia tables
and contains a front end for Erlang tracing.
</p>
</section>
<section>
<title>Getting Started</title>
<p>Run Observer from a standalone node to minimize the impact of the
system being observed.
</p>
<p><em>Example:</em></p>
<pre>
% <input>erl -sname observer -hidden -setcookie MyCookie -run observer</input></pre>
<p>
Select the node to observe with menu <em>Nodes</em>.
Menu <em>View > Refresh interval</em> controls how often
the view is to be updated.
The refresh interval is set per viewer so you can
have different settings for each viewer. To minimize the system
impact, only the active viewer is updated. Other views are updated
when activated.
</p>
<p>The mouse buttons behave as expected. Use left-click
to select objects, right-click to get a menu with the most used
options, and double-click to display information about the
selected object. In most viewers with many columns, you can change
the sort order by left-clicking the column header.
</p>
</section>
<section>
<title>System Tab</title>
<p>Tab <em>System</em> displays general information about the active Erlang node
and its runtime system, such as build configuration, system capabilities, and
overall use statistics.
</p>
</section>
<section>
<title>Load Charts Tab</title>
<p>Tab <em>Load Charts</em> displays graphs of the current resource use on
the active Erlang node.</p>
<p>Graph <c>Scheduler Utilization</c> shows scheduler use per scheduler,
where each scheduler use has a unique color.</p>
<p>Graph <c>Memory Usage</c> shows the total memory use and per memory category
use, where each category has a unique color. The categories are as
follows:</p>
<taglist>
<tag><c>Total</c></tag>
<item><p>The sum of all memory categories.</p></item>
<tag><c>Processes</c></tag>
<item><p>The sum of all process memory used.</p></item>
<tag><c>Atom</c></tag>
<item><p>The size used by the atom table.</p></item>
<tag><c>Binary</c></tag>
<item><p>The sum of all off-heap binaries allocated.</p></item>
<tag><c>Code</c></tag>
<item><p>The memory allocated for code storage.</p></item>
<tag><c>Ets</c></tag>
<item><p>The used memory for all ETS tables.</p></item>
</taglist>
<p>Graph <c>IO Usage</c> shows the current I/O load on the system.</p>
</section>
<section>
<title>Memory Allocators Tab</title>
<p>Tab <em>Memory Allocators</em> displays detailed information of the carrier
size and current memory carriers. For details about memory carriers,
see module
<seealso marker="erts:erts_alloc"><c>erts_alloc</c></seealso>
in application ERTS.</p>
<p>The <c>Max Carrier size</c> column shows the maximum value seen by observer
since the last node change or since the start of the application, i.e. switching
nodes will reset the max column. Values are sampled so higher values may have
existed than what is shown.
</p>
</section>
<section>
<title>Applications Tab</title>
<p>Tab <em>Applications</em> presents application information.
Select an application in the left list to display its supervisor
tree. The right-click options in the tree are as follows:
</p>
<taglist>
<tag>Process info</tag>
<item><p>Opens a detailed information window on the selected process,
including the following:</p>
<taglist>
<tag>Process Information</tag>
<item><p>Shows the process information.</p></item>
<tag>Messages</tag>
<item><p>Shows the process messages.</p></item>
<tag>Dictionary</tag>
<item><p>Shows the process dictionary.</p></item>
<tag>Stack Trace</tag>
<item><p>Shows the process current stack trace.</p></item>
<tag>State</tag>
<item><p>Shows the process state.</p></item>
<tag>Log</tag>
<item><p>If enabled and available, shows the process SASL
log entries.</p></item>
</taglist>
</item>
<tag>Trace process</tag>
<item><p>Adds the selected process identifier to tab <em>Trace Overview</em>
plus the node that the process resides on.</p></item>
<tag>Trace named process</tag>
<item><p>Adds the registered name of the process. This can be useful when tracing on
many nodes, as processes with that name are then traced on all traced nodes.</p></item>
<tag>Trace process tree</tag>
<item><p>Adds the selected process and all processes below,
right of it, to tab <em>Trace Overview</em>.</p></item>
<tag>Trace named process tree</tag>
<item><p>Adds the selected process and all processes below,
right of it, to tab <em>Trace Overview</em>.</p></item>
</taglist>
</section>
<section>
<title>Processes Tab</title>
<p>Tab <em>Processes</em> lists process information in columns.
For each process the following information is displayed:
</p>
<taglist>
<tag>Pid</tag>
<item><p>The process identifier.</p></item>
<tag>Reds</tag>
<item><p>The number of reductions executed on the process.
This can be presented as accumulated values or as values since the last update.</p></item>
<tag>Memory</tag>
<item><p>The size of the process, in bytes, obtained by a
call to <c>process_info(Pid,memory)</c>.</p></item>
<tag>MsgQ</tag>
<item><p>The length of the message queue for the process.</p></item>
</taglist>
<p>Option <em>Process info</em> opens a detailed information window on the process under the mouse pointer,
including the following:</p>
<taglist>
<tag>Process Information</tag>
<item><p>Shows the process information.</p></item>
<tag>Messages</tag>
<item><p>Shows the process messages.</p></item>
<tag>Dictionary</tag>
<item><p>Shows the process dictionary.</p></item>
<tag>Stack Trace</tag>
<item><p>Shows the process current stack trace.</p></item>
<tag>State</tag>
<item><p>Shows the process state.</p></item>
<tag>Log</tag>
<item><p>If enabled and available, shows the process SASL log entries.</p></item>
</taglist>
<note>
<p><em>Log</em> requires application SASL to be started on the observed node,
with <c>log_mf_h</c> as log handler.
The Observed node must be Erlang/OTP R16B02 or higher.
The <c>rb</c> server must not be started on the observed node when clicking menu
<em>Log > Toggle log view</em>. The <c>rb</c> server is stopped on the observed node
when exiting or changing the observed node.
</p>
</note>
<p>Option <em>Trace selected processes</em> adds the selected process identifiers to tab
<em>Trace Overview</em> plus the node that the processes reside on.
</p>
<p>Option <em>Trace selected processes by name</em> adds the registered name of the processes. This can be
useful when tracing is done on many nodes, as processes with that name are then traced on
all traced nodes.</p>
<p>Option <em>Kill process</em> brutally kills the processes under
the mouse pointer by sending an exit signal with
reason <c>kill</c>.</p>
</section>
<section>
<title>Ports Tab</title>
<p>Tab <em>Ports</em> lists port information in columns.
For each port the following information is displayed:
</p>
<taglist>
<tag>Id</tag>
<item><p>The port identifier.</p></item>
<tag>Connected</tag>
<item><p>The process identifier for the process that owns the
port.</p></item>
<tag>Name</tag>
<item><p>The registered name of the port, if any.</p></item>
<tag>Controls</tag>
<item><p>The name of the command set by <seealso marker="erts:erlang#open_port-2"><c>erlang:open_port/2</c></seealso>.</p></item>
<tag>Slot</tag>
<item><p>The internal index of the port.</p></item>
</taglist>
<p>Option <em>Port info</em> opens a detailed information window
for the port under the mouse pointer. In addition to the
information above, it also shows links and monitors.</p>
<p>Option <em>Trace selected ports</em> adds the selected port
identifiers, and the nodes that the ports reside on,
to tab <em>Trace Overview</em>.</p>
<p>Option <em>Trace selected ports by name</em> adds the
registered name of the port to tab <em>Trace Overview</em>. This
can be useful when tracing is done on many nodes, as ports with
that name are then traced on all traced nodes.</p>
<p>Option <em>Close</em>
executes <seealso marker="erts:erlang#port_close-1"><c>erlang:port_close/1</c></seealso>
on the port under the mouse pointer.</p>
</section>
<section>
<title>Table Viewer Tab</title>
<p>Tab <em>Table Viewer</em> lists tables. By default, ETS tables
are displayed whereas unreadable private ETS tables and tables created by OTP
applications are not diplayed. Use menu <em>View</em> to view "system"
ETS tables, unreadable ETS tables, or Mnesia tables.
</p>
<p>Double-click to view the table content, or right-click and
select option <em>Show Table Content</em>. To view table
information, select the table and activate menu <em>View >
Table information</em>, or right-click and select option <em>Table
info</em>.</p>
<p>You can use <seealso marker="stdlib:re">regular
expressions</seealso> and search for objects, and edit or delete them.
</p>
</section>
<section>
<title>Trace Overview Tab</title>
<p>Tab <em>Trace Overview</em> handles tracing. Trace by selecting
the processes or ports to be traced and how to trace them. For
processes, you can trace messages, function calls, scheduling,
garbage collections, and process-related events such
as <c>spawn</c>, <c>exit</c>, and many others. For ports, you can
trace messages, scheduling and port-related events.
</p>
<p>To trace function calls, you also need to set up
<em>trace patterns</em>. Trace patterns select the function calls
to be traced. The number of traced function calls can be
further reduced with <em>match specifications</em>. Match
specifications can also be used to trigger more information
in the trace messages.
</p>
<p>You can also set match specifications on messages. By default,
if tracing messages, all messages sent and/or received by the
process or port are traced. Match specifications can be used to
reduce the number of traced messages and/or to trigger more
information in the trace messages.</p>
<note><p>Trace patterns only apply to the traced processes and
ports.</p></note>
<p>
Processes are added from the <em>Applications</em>
or <em>Processes</em> tabs. Ports are added from
the <em>Ports</em> tab. A special <em>new</em> identifier,
meaning all processes, or ports, started after trace start, can
be added with buttons <em>Add 'new' Processes</em> and <em>Add
'new' Ports</em>, respecively.
</p>
<p>
When adding processes or ports, a window with trace options is
displayed. The chosen options are set for the selected
processes/ports. To change the options, right-click the process
or port and select <em>Edit process options</em>. To remove a
process or port from the list, right-click and select <em>Remove
process</em> or <em>Remove port</em>, respectively.
</p>
<p>
Processes and ports added by process/port identifiers add the
nodes these processes/ports reside on in the node list. More
nodes can be added by clicking button <em>Add Nodes</em>, or by
right-clicking in the <em>Nodes</em> list and select <em>Add
Nodes</em>. To remove nodes, select them, then right-click and
choose <em>Remove nodes</em>.
</p>
<p>
If function calls are traced, trace patterns must be added by clicking button
<em>Add Trace Pattern</em>. Select a module, function(s), and a match specification.
If no functions are selected, all functions in the module are traced.</p>
<p>
Trace patterns can also be added for traced messages. Click
button <em>Add Trace Pattern</em> and select <em>Messages
sent</em> or <em>Messages received</em>, and a match
specification.
</p>
<p>
A few basic match specifications are provided in the tool, and
you can provide your own match specifications. The syntax of match
specifications is described in the <seealso
marker="erts:match_spec"><c>ERTS User's Guide</c></seealso>. To simplify
the writing of a match specification, they can also be written as
<c>fun/1</c>. For details, see module
<seealso marker="stdlib:ms_transform">ms_transform</seealso>
in application STDLIB.
</p>
<p>Click button <em>Start Trace</em> to start the trace.
By default, trace output is written to a new window. Tracing is stopped
when the window is closed, or when clicking button <em>Stop Trace</em>.
Trace output can be changed with menu <em>Options > Output</em>.
The trace settings, including match specifications, can be saved to,
or loaded from, a file.
</p>
<p>For details about tracing, see module <seealso
marker="runtime_tools:dbg">dbg</seealso> in application Runtime_Tools
and in section "Match specifications in Erlang" in
<seealso marker="erts:match_spec"><c>ERTS User's Guide</c></seealso>
and in module
<seealso marker="stdlib:ms_transform"><c>ms_transform</c></seealso>
in application STDLIB.
</p>
</section>
</chapter>
