20032013 Ericsson AB. All Rights Reserved. The contents of this file are subject to the Erlang Public License, Version 1.1, (the "License"); you may not use this file except in compliance with the License. You should have received a copy of the Erlang Public License along with this software. If not, it can be retrieved online at http://www.erlang.org/. Software distributed under the License is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License for the specific language governing rights and limitations under the License. Crashdump Viewer crashdump_ug.xml
Introduction

The Crashdump Viewer is a WxWidgets based tool for browsing Erlang crashdumps.

Getting Started

The easiest way to start Crashdump Viewer is to use the provided shell script named cdv with the full path to the erlang crashdump as an argument. The script can be found in the priv directory of the observer application. This starts the Crashdump Viewer GUI and loads the given file. If no file name is given, a file dialog will be opened where the file can be selected.

Under Windows the batch file cdv.bat can be used.

It is also possible to start the Crashdump Viewer from within an erlang node by calling crashdump_viewer:start/0 or crashdump_viewer:start/1.

The graphical interface

The main window is opened when Crashdump Viewer has loaded a crashdump. It contains a title bar, a menu bar, a number of information panels and a status bar.

The title bar shows the name of the currently loaded crashdump.

The menu bar contains a File menu and a Help menu. From the File menu a new crashdump can be loaded or the tool can be terminated. From the Help menu this user's guide and the chapter "How to interpret the Erlang crash dumps" from the user's guide for Erlang runtime system can be opened. "How to interpret the Erlang crash dumps" describes the raw crashdumps in detail. Here you will also find information about each single field in the different information pages. This document can also be found directly in the OTP online documentation, via the Erlang runtime system user's guide.

The status bar at the bottom of the window will show a warning if the currently loaded dump is truncated.

The centre area of the main window contains the information panels. Each panel displays information about a specific item or a list of items. A panel is selected by clicking the title of the tab.

From panels that display lists of items, for example the Processes- or the Ports panel, a new window with further information can be opened by double clicking a row or by right clicking the row and selecting an item from the drop down menu. The new window is called a detail window. Detail windows can be opened for processes, ports, nodes and modules.

The various information shown in a detail window might contain links to processes or ports. Clicking one of these links will open the detail window for the process or port in question. If the process or port resided on a remote node, there will be no information available. Clicking the link will then pop up a dialog where you can choose whether to open the detail window for the remote node or not.

Some of the panels contain a left hand menu where sub items of the panel's information area can be selected. Click on one of the rows, and the information will be displayed in the right hand information area.

Data content

Each panel in the main window contains an information page. If no information is found for an item, the page will be empty. The reason for not finding any information about an item can be that the dump is truncated, that it is a dump from an old OTP release in which this item was not written or that the item simply wasn't present in the system at the point of failure.

If the dump was truncated, a warning is displayed in the status bar of the main window.

Even if some information about an item exists, there might be empty fields if the dump originates from an old OTP release.

The value "-1" in any field means "unknown", and in most cases it means that the dump was truncated somewhere around this field.

The sections below describe some of the fields in the different information panels. These are fields that do not exist in the raw crashdump, or in some way differ from the fields in the raw crashdump. Details about other fields can be found in the user's guide for the Erlang runtime system, in the chapter "How to interpret the Erlang crash dumps". That chapter can also be opened from the Help menu in the Crashdump Viewer's main window, and there are also direct links from the specific sections below to related information in "How to interpret the Erlang crash dumps".

General information

The General information panel shows a short overview of the dump.

The following fields are not described in the Erlang runtime system user's guide:

Crashdump created on Time of failure. Memory allocated The total number of bytes allocated, equivalent to c:memory(total). Memory maximum The maximum number of bytes that has been allocated during the lifetime of the originating node. This will only be shown if the Erlang runtime system was run instrumented. Atoms If available in the dump, this is the total number of atoms in the atom table. If the size of the atom table is not available, the number of atoms visible in the dump is presented. Processes, ETS tables and Funs The number of processes, ETS tables and funs visible in the dump.

More...

Processes

The Processes panel shows a list of all processes found in the crashdump, including some short information about each process. By default the processes are sorted by their pids. To sort by other topic, click the desired column heading.

The Memory column shows the 'Memory' field which was added to crashdumps in R16B01. This is the total amount of memory used by the process. For crashdumps from earlier releases, this column shows the 'Stack+heap' field. The value shown is always in bytes.

To view detailed information about a specific process, double click the row in the list or right click the row and select "Properties for <pid>".

More...

Ports

The Ports panel is similar to the Processes panel, except it lists all ports found in the crashdump.

To see more details about a specific port, dobule click the row or right click it and select "Properties for <port>". From the right click menu you can also select "Properties for <pid>", where <pid> is the process connected to the port.

More...

ETS tables

The ETS Tables panel shows all ETS table information found in the dump. The 'Id' is the same as the 'Table' field found in the raw crashdump, and 'Memory' is the 'Words' field from the raw crashdump translated into bytes. 'Type' is the type of table, and it can be either "hash" or "tree". For tree tables there will be no value in the 'Bucket' field.

To open the detailed information page about the owner process of an ETS table, right click the row and select "Properties for <pid>".

Double clicking a row in the ETS Tables panel has no effect.

From the left hand menu you can also select to see internal ETS tables.

More...

Timers

The Timers panel shows all timer information found in the dump.

To open the detailed information page about the owner process of a timer, right click the row and select "Properties for <pid>".

Double clicking a row in the Timers panel has no effect.

More...

Funs

The Funs panel shows all Fun information found in the dump.

To open the detailed information page about the module to which the fun belongs, right click the row and select "Properties for <mod>".

Double clicking a row in the Funs panel has no effect.

More...

Atoms

The Atoms panel lists all atoms found in the dump. By default the atoms are sorted in creation order from first to last. This is opposite of the raw crashdump where atoms are listed from last to first, meaning that if the dump was truncated in the middle of the atom list only the last created atoms will be seen in the Atoms panel.

More...

Nodes

The Nodes panel shows a list of all external erlang nodes which are referenced from the crashdump.

If the page is empty it either means that the crashed node was not distributed, that it was distributed but had no references to other nodes or that the dump was truncated.

If the node was distributed, all referenced nodes are shown. The column named Connection type shows if the node is visible, hidden or not connected. Visible nodes are alive nodes with a living connection to the originating node. Hidden nodes are the same as visible nodes, except they are started with the -hidden flag. Not connected nodes are nodes that are not connected to the originating node anymore, but references (i.e. process or port identifiers) exist.

To see more detailed information about a node, double click the row or right click the row and select "Properties for node <node>". From the right click menu you can also select "Properties for <port>", to open the detailed information window for the controlling port.

In the detailed information window for a node, any exsisting links and monitors between processes on the originating node and the connected node are shown. Extra Info may contain debug information (i.e. special information written if the emulator is debug compiled) or error information.

More...

Loaded modules

The Modules panel lists all modules that were loaded on the originating node, and the current size of the code. If old code exsits, the old size is also shown.

To see detailed information about a specific module, double click the row or right click it and select "Properties for <mod>".

More...

Memory

The Memory panel shows memory and allocator information. From the left hand menu you can select:

Memory More... Allocator Summary - this page presents a summary of values from all allocators below. <Allocator> - one entry per allocator More... Allocated Areas More...

Internal tables

On the Internal Tables panel you can choose from the left hand menu to see hash tables or index tables.

More...