systemd/man/systemd-coredump.xml
2022-01-12 16:05:59 +01:00

437 lines
22 KiB
XML

<?xml version='1.0'?>
<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
"http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd">
<!-- SPDX-License-Identifier: LGPL-2.1-or-later -->
<refentry id="systemd-coredump" conditional='ENABLE_COREDUMP'
xmlns:xi="http://www.w3.org/2001/XInclude">
<refentryinfo>
<title>systemd-coredump</title>
<productname>systemd</productname>
</refentryinfo>
<refmeta>
<refentrytitle>systemd-coredump</refentrytitle>
<manvolnum>8</manvolnum>
</refmeta>
<refnamediv>
<refname>systemd-coredump</refname>
<refname>systemd-coredump.socket</refname>
<refname>systemd-coredump@.service</refname>
<refpurpose>Acquire, save and process core dumps</refpurpose>
</refnamediv>
<refsynopsisdiv>
<para><filename>/usr/lib/systemd/systemd-coredump</filename></para>
<para><filename>/usr/lib/systemd/systemd-coredump</filename> <option>--backtrace</option></para>
<para><filename>systemd-coredump@.service</filename></para>
<para><filename>systemd-coredump.socket</filename></para>
</refsynopsisdiv>
<refsect1>
<title>Description</title>
<para><filename>systemd-coredump@.service</filename> is a system service to process core dumps. It will
log a summary of the event to
<citerefentry><refentrytitle>systemd-journald.service</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
including information about the process identifier, owner, the signal that killed the process, and the
stack trace if possible. It may also save the core dump for later processing. See the "Information about
the crashed process" section below.</para>
<para>The behavior of a specific program upon reception of a signal is governed by a few
factors which are described in detail in
<citerefentry project='man-pages'><refentrytitle>core</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
In particular, the core dump will only be processed when the related resource limits are sufficient.
</para>
<para>Core dumps can be written to the journal or saved as a file. In both cases, they can be retrieved
for further processing, for example in
<citerefentry project='man-pages'><refentrytitle>gdb</refentrytitle><manvolnum>1</manvolnum></citerefentry>.
See <citerefentry><refentrytitle>coredumpctl</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
in particular the <command>list</command> and <command>debug</command> verbs.</para>
<para>By default, <command>systemd-coredump</command> will log the core dump to the journal, including a
backtrace if possible, and store the core dump (an image of the memory contents of the process) itself in
an external file in <filename>/var/lib/systemd/coredump</filename>. These core dumps are deleted after a
few days by default; see <filename>/usr/lib/tmpfiles.d/systemd.conf</filename> for details. Note that the
removal of core files from the file system and the purging of journal entries are independent, and the
core file may be present without the journal entry, and journal entries may point to since-removed core
files. Some metadata is attached to core files in the form of extended attributes, so the core files are
useful for some purposes even without the full metadata available in the journal entry.</para>
<refsect2>
<title>Invocation of <command>systemd-coredump</command></title>
<para>The <command>systemd-coredump</command> executable does the actual work. It is invoked twice:
once as the handler by the kernel, and the second time in the
<filename>systemd-coredump@.service</filename> to actually write the data to the journal and process
and save the core file.</para>
<para>When the kernel invokes <command>systemd-coredump</command> to handle a core dump, it runs in
privileged mode, and will connect to the socket created by the
<filename>systemd-coredump.socket</filename> unit, which in turn will spawn an unprivileged
<filename>systemd-coredump@.service</filename> instance to process the core dump. Hence
<filename>systemd-coredump.socket</filename> and <filename>systemd-coredump@.service</filename> are
helper units which do the actual processing of core dumps and are subject to normal service
management.</para>
<para>It is also possible to invoke <command>systemd-coredump</command> with
<option>--backtrace</option> option. In this case, <command>systemd-coredump</command> expects a
journal entry in the journal
<ulink url="https://systemd.io/JOURNAL_EXPORT_FORMATS#journal-export-format">Journal Export Format</ulink>
on standard input. The entry should contain a <varname>MESSAGE=</varname> field and any additional
metadata fields the caller deems reasonable. <command>systemd-coredump</command> will append additional
metadata fields in the same way it does for core dumps received from the kernel. In this mode, no core
dump is stored in the journal.</para>
</refsect2>
</refsect1>
<refsect1>
<title>Configuration</title>
<para>For programs started by <command>systemd</command>, process resource limits can be set by directive
<varname>LimitCORE=</varname>, see
<citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
<para>In order to be used by the kernel to handle core dumps,
<command>systemd-coredump</command> must be configured in
<citerefentry project='man-pages'><refentrytitle>sysctl</refentrytitle><manvolnum>8</manvolnum></citerefentry>
parameter <varname>kernel.core_pattern</varname>. The syntax of this parameter is explained in
<citerefentry project='man-pages'><refentrytitle>core</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
systemd installs the file <filename>/usr/lib/sysctl.d/50-coredump.conf</filename> which configures
<varname>kernel.core_pattern</varname> accordingly. This file may be masked or overridden to use a different
setting following normal
<citerefentry><refentrytitle>sysctl.d</refentrytitle><manvolnum>5</manvolnum></citerefentry>
rules. If the sysctl configuration is modified, it must be updated in the kernel before it
takes effect, see
<citerefentry project='man-pages'><refentrytitle>sysctl</refentrytitle><manvolnum>8</manvolnum></citerefentry>
and
<citerefentry><refentrytitle>systemd-sysctl</refentrytitle><manvolnum>8</manvolnum></citerefentry>.
</para>
<para>In order to be used in the <option>--backtrace</option> mode, an appropriate backtrace
handler must be installed on the sender side. For example, in case of
<citerefentry project='die-net'><refentrytitle>python</refentrytitle><manvolnum>1</manvolnum></citerefentry>, this
means a <varname>sys.excepthook</varname> must be installed, see
<ulink url="https://github.com/systemd/systemd-coredump-python">systemd-coredump-python</ulink>.
</para>
<para>The behavior of <command>systemd-coredump</command> itself is configured through the configuration file
<filename>/etc/systemd/coredump.conf</filename> and corresponding snippets
<filename>/etc/systemd/coredump.conf.d/*.conf</filename>, see
<citerefentry><refentrytitle>coredump.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>. A new
instance of <command>systemd-coredump</command> is invoked upon receiving every core dump. Therefore, changes
in these files will take effect the next time a core dump is received.</para>
<para>Resources used by core dump files are restricted in two ways. Parameters like maximum size of acquired
core dumps and files can be set in files <filename>/etc/systemd/coredump.conf</filename> and snippets mentioned
above. In addition the storage time of core dump files is restricted by <command>systemd-tmpfiles</command>,
corresponding settings are by default in <filename>/usr/lib/tmpfiles.d/systemd.conf</filename>. The default is
to delete core dumps after a few days; see the above file for details.</para>
<refsect2>
<title>Disabling coredump processing</title>
<para>To disable potentially resource-intensive processing by <command>systemd-coredump</command>, set
<programlisting>Storage=none ProcessSizeMax=0</programlisting> in
<citerefentry><refentrytitle>coredump.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
</refsect2>
</refsect1>
<refsect1>
<title>Information about the crashed process</title>
<para><citerefentry><refentrytitle>coredumpctl</refentrytitle><manvolnum>1</manvolnum></citerefentry> can
be used to retrieve saved core dumps independently of their location, to display information, and to
process them e.g. by passing to the GNU debugger (gdb).</para>
<para>Data stored in the journal can be also viewed with
<citerefentry><refentrytitle>journalctl</refentrytitle><manvolnum>1</manvolnum></citerefentry> as usual
(or from any other process, using the
<citerefentry><refentrytitle>sd-journal</refentrytitle><manvolnum>3</manvolnum></citerefentry> API).
The relevant messages have <constant>MESSAGE_ID=fc2e22bc6ee647b6b90729ab34a250b1</constant>:</para>
<programlisting>$ journalctl MESSAGE_ID=fc2e22bc6ee647b6b90729ab34a250b1 -o verbose
MESSAGE_ID=fc2e22bc6ee647b6b90729ab34a250b1
COREDUMP_PID=552351
COREDUMP_UID=1000
COREDUMP_GID=1000
COREDUMP_SIGNAL_NAME=SIGSEGV
COREDUMP_SIGNAL=11
COREDUMP_TIMESTAMP=1614342930000000
COREDUMP_COMM=Web Content
COREDUMP_EXE=/usr/lib64/firefox/firefox
COREDUMP_USER_UNIT=app-gnome-firefox-552136.scope
COREDUMP_CMDLINE=/usr/lib64/firefox/firefox -contentproc -childID 5 -isForBrowser …
COREDUMP_CGROUP=/user.slice/user-1000.slice/user@1000.service/app.slice/app-….scope
COREDUMP_FILENAME=/var/lib/systemd/coredump/core.Web….552351.….zst
</programlisting>
<para>The following fields are saved (if known) with the journal entry</para>
<variablelist class='journal-directives'>
<varlistentry>
<term><varname>COREDUMP_UID=</varname></term>
<term><varname>COREDUMP_PID=</varname></term>
<term><varname>COREDUMP_GID=</varname></term>
<listitem><para>The process number (PID), owner user number (UID), and group number (GID) of the
crashed process.</para>
<para>When the crashed process was part of a container (or in a process or user namespace in
general), those are the values as seen <emphasis>outside</emphasis>, in the namespace where
<filename>systemd-coredump</filename> is running.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_TIMESTAMP=</varname></term>
<listitem><para>The time of the crash as reported by the kernel (in µs since the epoch).</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_RLIMIT=</varname></term>
<listitem><para>The core file size soft resource limit, see
<citerefentry project='man-pages'><refentrytitle>getrlimit</refentrytitle><manvolnum>2</manvolnum></citerefentry>.
</para></listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_UNIT=</varname></term>
<term><varname>COREDUMP_SLICE=</varname></term>
<listitem><para>The system unit and slice names.</para>
<para>When the crashed process was in container, those are the units names
<emphasis>outside</emphasis>, in the main system manager.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_CGROUP=</varname></term>
<listitem><para>Control group information in the format used in
<filename>/proc/self/cgroup</filename>. On systems with the unified cgroup hierarchy, this is a
single path prefixed with <literal>0::</literal>, and multiple paths prefixed with controller numbers
on legacy systems.</para>
<para>When the crashed process was in a container, this is the full path, as seen outside of the
container.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_OWNER_UID=</varname></term>
<term><varname>COREDUMP_USER_UNIT=</varname></term>
<listitem><para>The numerical UID of the user owning the login session or systemd user unit of the
crashed process, and the user manager unit. Both fields are only present for user processes.
</para>
<para>When the crashed process was in container, those are the values <emphasis>outside</emphasis>,
in the main system.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_SIGNAL_NAME=</varname></term>
<term><varname>COREDUMP_SIGNAL=</varname></term>
<listitem><para>The terminating signal name (with the <literal>SIG</literal> prefix
<footnote><para><citerefentry
project='man-pages'><refentrytitle>kill</refentrytitle><manvolnum>1</manvolnum></citerefentry>
expects signal names <emphasis>without</emphasis> the prefix; <citerefentry
project='man-pages'><refentrytitle>kill</refentrytitle><manvolnum>2</manvolnum></citerefentry> uses
the prefix; all systemd tools accept signal names both with and without the prefix.
</para></footnote>) and numerical value. (Both are included because signal numbers vary by
architecture.)</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_CWD=</varname></term>
<term><varname>COREDUMP_ROOT=</varname></term>
<listitem><para>The current working directory and root directory of the crashed process.</para>
<para>When the crashed process is in a container, those paths are relative to the root of the
container's mount namespace.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_OPEN_FDS=</varname></term>
<listitem><para>Information about open file descriptors, in the following format:</para>
<programlisting><replaceable>fd</replaceable>:<replaceable>/path/to/file</replaceable>
pos: ...
flags: ...
...
<replaceable>fd</replaceable>:<replaceable>/path/to/file</replaceable>
pos: ...
flags: ...
...
</programlisting>
<para>The first line contains the file descriptor number <replaceable>fd</replaceable> and the path,
while subsequent lines show the contents of
<filename>/proc/<replaceable>pid</replaceable>/fdinfo/<replaceable>fd</replaceable></filename>.
</para></listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_EXE=</varname></term>
<listitem><para>The destination of the <filename>/proc/<replaceable>pid</replaceable>/exe</filename>
symlink.</para>
<para>When the crashed process is in a container, that path is relative to the root of the
container's mount namespace.</para></listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_COMM=</varname></term>
<term><varname>COREDUMP_PROC_STATUS=</varname></term>
<term><varname>COREDUMP_PROC_MAPS=</varname></term>
<term><varname>COREDUMP_PROC_LIMITS=</varname></term>
<term><varname>COREDUMP_PROC_MOUNTINFO=</varname></term>
<term><varname>COREDUMP_ENVIRON=</varname></term>
<listitem><para>Fields that map the per-process entries in the <filename>/proc/</filename>
filesystem: <filename>/proc/<replaceable>pid</replaceable>/comm</filename> (the command name
associated with the process), <filename>/proc/<replaceable>pid</replaceable>/exe</filename> (the
filename of the executed command), <filename>/proc/<replaceable>pid</replaceable>/status</filename>
(various metadata about the process), <filename>/proc/<replaceable>pid</replaceable>/maps</filename>
(memory regions visible to the process and their access permissions),
<filename>/proc/<replaceable>pid</replaceable>/limits</filename> (the soft and hard resource limits),
<filename>/proc/<replaceable>pid</replaceable>/mountinfo</filename> (mount points in the process's
mount namespace), <filename>/proc/<replaceable>pid</replaceable>/environ</filename>
(the environment block of the crashed process).</para>
<para>See
<citerefentry project='man-pages'><refentrytitle>proc</refentrytitle><manvolnum>5</manvolnum></citerefentry>
for more information.</para></listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_HOSTNAME=</varname></term>
<listitem><para>The system hostname.</para>
<para>When the crashed process was in container, this is the container hostname.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_CONTAINER_CMDLINE=</varname></term>
<listitem><para>For processes running in a container, the commandline of the process spawning the
container (the first parent process with a different mount namespace).</para></listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP=</varname></term>
<listitem><para>When the core is stored in the journal, the core image itself.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_FILENAME=</varname></term>
<listitem><para>When the core is stored externally, the path to the core file.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_TRUNCATED=</varname></term>
<listitem><para>Set to <literal>1</literal> when the saved coredump was truncated. (A partial core
image may still be processed by some tools, though obviously not all information is available.)
</para></listitem>
</varlistentry>
<varlistentry>
<term><varname>COREDUMP_PACKAGE_NAME=</varname></term>
<term><varname>COREDUMP_PACKAGE_VERSION=</varname></term>
<term><varname>COREDUMP_PACKAGE_JSON=</varname></term>
<listitem><para>If the executable contained .package metadata ELF notes, they will be
parsed and attached. The <varname>package</varname> and <varname>packageVersion</varname>
of the 'main' ELF module (ie: the executable) will be appended individually. The
JSON-formatted content of all modules will be appended as a single JSON object, each with
the module name as the key. For more information about this metadata format and content, see
<ulink url="https://systemd.io/COREDUMP_PACKAGE_METADATA/">the coredump metadata spec</ulink>.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>MESSAGE=</varname></term>
<listitem><para>The message generated by <command>systemd-coredump</command> that includes the
backtrace if it was successfully generated. When <command>systemd-coredump</command> is invoked with
<option>--backtrace</option>, this field is provided by the caller.</para></listitem>
</varlistentry>
</variablelist>
<para>Various other fields exist in the journal entry, but pertain to the logging process,
i.e. <command>systemd-coredump</command>, not the crashed process. See
<citerefentry><refentrytitle>systemd.journal-fields</refentrytitle><manvolnum>7</manvolnum></citerefentry>.
</para>
<para>The following fields are saved (if known) with the external file listed in
<varname>COREDUMP_FILENAME=</varname> as extended attributes:</para>
<variablelist>
<varlistentry>
<term><varname>user.coredump.pid</varname></term>
<term><varname>user.coredump.uid</varname></term>
<term><varname>user.coredump.gid</varname></term>
<term><varname>user.coredump.signal</varname></term>
<term><varname>user.coredump.timestamp</varname></term>
<term><varname>user.coredump.rlimit</varname></term>
<term><varname>user.coredump.hostname</varname></term>
<term><varname>user.coredump.comm</varname></term>
<term><varname>user.coredump.exe</varname></term>
<listitem><para>Those are the same as <varname>COREDUMP_PID=</varname>,
<varname>COREDUMP_UID=</varname>, <varname>COREDUMP_GID=</varname>,
<varname>COREDUMP_SIGNAL=</varname>, <varname>COREDUMP_TIMESTAMP=</varname>,
<varname>COREDUMP_RLIMIT=</varname>, <varname>COREDUMP_HOSTNAME=</varname>,
<varname>COREDUMP_COMM=</varname>, and <varname>COREDUMP_EXE=</varname>, described above.
</para></listitem>
</varlistentry>
</variablelist>
<para>Those can be viewed using
<citerefentry project='man-pages'><refentrytitle>getfattr</refentrytitle><manvolnum>1</manvolnum></citerefentry>.
For the core file described in the journal entry shown above:
<programlisting>$ getfattr --absolute-names -d /var/lib/systemd/coredump/core.Web….552351.….zst
# file: /var/lib/systemd/coredump/core.Web….552351.….zst
user.coredump.pid="552351"
user.coredump.uid="1000"
user.coredump.gid="1000"
user.coredump.signal="11"
user.coredump.timestamp="1614342930000000"
user.coredump.comm="Web Content"
user.coredump.exe="/usr/lib64/firefox/firefox"
</programlisting>
</para>
</refsect1>
<refsect1>
<title>See Also</title>
<para>
<citerefentry><refentrytitle>coredump.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
<citerefentry><refentrytitle>coredumpctl</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
<citerefentry><refentrytitle>systemd-journald.service</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
<citerefentry><refentrytitle>systemd-tmpfiles</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
<citerefentry project='man-pages'><refentrytitle>core</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
<citerefentry><refentrytitle>sysctl.d</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
<citerefentry><refentrytitle>systemd-sysctl.service</refentrytitle><manvolnum>8</manvolnum></citerefentry>.
</para>
</refsect1>
</refentry>