Files
Eric BoucherandZbigniew Jędrzejewski-Szmek 34127b7c4c cgtop: make root row default to cgroup with '/' to toggle
The root row in systemd-cgtop previously displayed whole-system RAM
usage read from /proc/meminfo (MemTotal - MemAvailable), not a cgroup
sum. This is a pragmatic fallback — the kernel intentionally does not
expose memory.current on the v2 root cgroup (see the existing
is_root_cgroup() comment about avoiding 'multiple truth about system
state'), so cgtop picked /proc/meminfo to fill the gap.

Nevertheless, that caued the '/' row to provide different information
than every other row in the table:
- All child rows: 'what is this cgroup charged for?' (anon + file
  + sock + shmem attributable to the cgroup).
- Root row: 'what is the whole system using?' (includes kernel
  slab, vmalloc, pagetables, zram mem_used, and anything not
  attributable to any cgroup).

On a typical desktop the root row can read ~10 GB while the sum of
direct children is ~5 GB, with no explanation visible in the tool
itself.

This commit:
  1. Introduces --root-memory= command line option.
  2. Flips the default for the root row to the sum of memory.current
     over its direct children.
  3. Adds a '/' interactive key that toggles between the two modes.
  4. Updates the interactive help string ('?' / 'h'), the man page,
     and the bash completion for the new option.

Inside a container (detect_container() > 0), is_root_cgroup() already
returns false, so none of this applies — the container's '/' still
reads memory.current from the namespaced cgroup as before.

Co-authored-by: Zbigniew Jędrzejewski-Szmek <zbyszek@amutable.com>
2026-10-09 11:19:05 +02:00

409 lines
16 KiB
XML

<?xml version='1.0'?>
<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
"http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd">
<!-- SPDX-License-Identifier: LGPL-2.1-or-later -->
<refentry id="systemd-cgtop"
xmlns:xi="http://www.w3.org/2001/XInclude">
<refentryinfo>
<title>systemd-cgtop</title>
<productname>systemd</productname>
</refentryinfo>
<refmeta>
<refentrytitle>systemd-cgtop</refentrytitle>
<manvolnum>1</manvolnum>
</refmeta>
<refnamediv>
<refname>systemd-cgtop</refname>
<refpurpose>Show top control groups by their resource usage</refpurpose>
</refnamediv>
<refsynopsisdiv>
<cmdsynopsis>
<command>systemd-cgtop</command>
<arg choice="opt" rep="repeat">OPTIONS</arg>
<arg choice="opt">GROUP</arg>
</cmdsynopsis>
</refsynopsisdiv>
<refsect1>
<title>Description</title>
<para><command>systemd-cgtop</command> shows the top control
groups of the local Linux control group hierarchy, ordered by
their CPU, memory, or disk I/O load. The display is refreshed in
regular intervals (by default every 1s), similar in style to
<citerefentry project='man-pages'><refentrytitle>top</refentrytitle><manvolnum>1</manvolnum></citerefentry>.
If a control group path is specified, shows only the services of
the specified control group.</para>
<para>If <command>systemd-cgtop</command> is not connected to a
tty, no column headers are printed and the default is to only run
one iteration. The <option>--iterations=</option> argument, if
given, is honored. This mode is suitable for scripting.</para>
<para>Resource usage is only accounted for control groups with the appropriate controllers turned on:
<literal>memory</literal> controller for memory usage and <literal>io</literal> controller for disk I/O consumption.
If resource monitoring for these resources is required, it is recommended to add the <varname>MemoryAccounting=1</varname>
and <varname>IOAccounting=1</varname> settings in the unit files in question. See
<citerefentry><refentrytitle>systemd.resource-control</refentrytitle><manvolnum>5</manvolnum></citerefentry>
for details.</para>
<para>The CPU load value can be between 0 and 100 times the number of
processors the system has. For example, if the system has 8 processors,
the CPU load value is going to be between 0% and 800%. The number of
processors can be found in <literal>/proc/cpuinfo</literal>.</para>
<para>To emphasize: unless <literal>MemoryAccounting=1</literal> and <literal>IOAccounting=1</literal>
are enabled for the services in question, no resource accounting will be available for system services
and the data shown by <command>systemd-cgtop</command> will be incomplete.</para>
</refsect1>
<refsect1>
<title>Options</title>
<para>The following options are understood:</para>
<variablelist>
<varlistentry>
<term><option>-p</option></term>
<term><option>--order=path</option></term>
<listitem><para>Order by control group
path name.</para></listitem>
</varlistentry>
<varlistentry>
<term><option>-t</option></term>
<term><option>--order=tasks</option></term>
<listitem><para>Order by number of tasks/processes in the control group.</para></listitem>
</varlistentry>
<varlistentry>
<term><option>-c</option></term>
<term><option>--order=cpu</option></term>
<listitem><para>Order by CPU load.</para></listitem>
</varlistentry>
<varlistentry>
<term><option>-m</option></term>
<term><option>--order=memory</option></term>
<listitem><para>Order by memory usage.</para></listitem>
</varlistentry>
<varlistentry>
<term><option>-i</option></term>
<term><option>--order=io</option></term>
<listitem><para>Order by disk I/O load.</para></listitem>
</varlistentry>
<varlistentry>
<term><option>-b</option></term>
<term><option>--batch</option></term>
<listitem><para>Run in "batch" mode: do not accept input and
run until the iteration limit set with
<option>--iterations=</option> is exhausted or until killed.
This mode could be useful for sending output from
<command>systemd-cgtop</command> to other programs or to a
file.</para>
<xi:include href="version-info.xml" xpointer="v188"/></listitem>
</varlistentry>
<varlistentry>
<term><option>-r</option></term>
<term><option>--raw</option></term>
<listitem><para>Format byte counts (as in memory usage and I/O metrics) and CPU time
with raw numeric values rather than human-readable
numbers.</para>
<xi:include href="version-info.xml" xpointer="v221"/></listitem>
</varlistentry>
<varlistentry>
<term><option>--cpu=percentage</option></term>
<term><option>--cpu=time</option></term>
<listitem><para>Controls whether the CPU usage is shown as
percentage or time. By default, the CPU usage is shown as
percentage. When the option is used without an argument, it is
equivalent to <option>--cpu=time</option>. This setting may also
be toggled at runtime by pressing the <keycap>%</keycap>
key.</para>
<xi:include href="version-info.xml" xpointer="v226"/></listitem>
</varlistentry>
<varlistentry>
<term><option>-P</option></term>
<listitem><para>Count only userspace processes instead of all
tasks. By default, all tasks are counted: each kernel thread
and each userspace thread individually. With this setting,
kernel threads are excluded from the count and each userspace
process only counts as one task, regardless of how many
threads it consists of. This setting may also be toggled at
runtime by pressing the <keycap>P</keycap> key. This option
may not be combined with
<option>-k</option>.</para>
<xi:include href="version-info.xml" xpointer="v227"/></listitem>
</varlistentry>
<varlistentry>
<term><option>-k</option></term>
<listitem><para>Count only userspace processes and kernel
threads instead of all tasks. By default, all tasks are
counted: each kernel thread and each userspace thread
individually. With this setting, kernel threads are included in
the count and each userspace process only counts as one task,
regardless of how many threads it consists of. This setting may
also be toggled at runtime by pressing the <keycap>k</keycap>
key. This option may not be combined with
<option>-P</option>.</para>
<xi:include href="version-info.xml" xpointer="v226"/></listitem>
</varlistentry>
<varlistentry>
<term><option>--recursive=</option></term>
<listitem><para>Controls whether the number of processes shown
for a control group shall include all processes that are
contained in any of the child control groups as well. Takes a
boolean argument, which defaults to <literal>yes</literal>. If
enabled, the processes in child control groups are included, if
disabled, only the processes in the control group itself are
counted. This setting may also be toggled at runtime by
pressing the <keycap>r</keycap> key. Note that this setting
only applies to process counting, i.e. when the
<option>-P</option> or <option>-k</option> options are
used. When all tasks are counted, setting this option to
<literal>no</literal> is rejected with an error, as task counting
is always recursive.</para>
<xi:include href="version-info.xml" xpointer="v226"/></listitem>
</varlistentry>
<varlistentry>
<term><option>-n</option></term>
<term><option>--iterations=</option></term>
<listitem><para>Perform only this many iterations. A value of
0 indicates that the program should run
indefinitely.</para>
<xi:include href="version-info.xml" xpointer="v188"/></listitem>
</varlistentry>
<varlistentry>
<term><option>-1</option></term>
<listitem><para>A shortcut for <option>--iterations=1</option>.</para>
<xi:include href="version-info.xml" xpointer="v238"/></listitem>
</varlistentry>
<varlistentry>
<term><option>-d</option></term>
<term><option>--delay=</option></term>
<listitem><para>Specify refresh delay in seconds (or if one of
<literal>ms</literal>, <literal>us</literal>,
<literal>min</literal> is specified as unit in this time
unit). This setting may also be increased and decreased at
runtime by pressing the <keycap>+</keycap> and
<keycap>-</keycap> keys.</para></listitem>
</varlistentry>
<varlistentry>
<term><option>--depth=</option></term>
<listitem><para>Maximum control group tree traversal depth.
Specifies how deep <command>systemd-cgtop</command> shall
traverse the control group hierarchies. If 0 is specified,
only the root group is monitored. For 1, only the first level
of control groups is monitored, and so on. Defaults to
3.</para></listitem>
</varlistentry>
<varlistentry>
<term><option>--root-memory=</option></term>
<listitem><para>Selects how the root control group row (<literal>/</literal>)
reports memory usage. Takes one of <literal>cgroup</literal> or
<literal>system</literal>. When <literal>cgroup</literal> (the default)
is used, the root row shows the sum of <filename>memory.current</filename>
across its direct children, which matches the accounting semantics of
every other row in the table. When <literal>system</literal> is used,
the root row shows whole-system RAM usage
(<literal>MemTotal - MemAvailable</literal> from
<filename>/proc/meminfo</filename>), which additionally accounts for
kernel slab, vmalloc, page tables, and other memory not attributable to
any cgroup. This setting may also be toggled at runtime by pressing the
<keycap>/</keycap> key.</para>
<xi:include href="version-info.xml" xpointer="v263"/></listitem>
</varlistentry>
<varlistentry>
<term><option>-M <replaceable>MACHINE</replaceable></option></term>
<term><option>--machine=<replaceable>MACHINE</replaceable></option></term>
<listitem><para>Limit control groups shown to the part
corresponding to the container
<replaceable>MACHINE</replaceable>.
This option may not be used when a control group path is specified.</para>
<xi:include href="version-info.xml" xpointer="v227"/></listitem>
</varlistentry>
<xi:include href="standard-options.xml" xpointer="help" />
<xi:include href="standard-options.xml" xpointer="version" />
</variablelist>
</refsect1>
<refsect1>
<title>Keys</title>
<para><command>systemd-cgtop</command> is an interactive tool and
may be controlled via user input using the following keys:</para>
<variablelist>
<varlistentry>
<term><keycap>h</keycap></term>
<listitem><para>Shows a short help text.</para></listitem>
</varlistentry>
<varlistentry>
<term><keycap function="space"/></term>
<listitem><para>Immediately refresh output.</para>
<xi:include href="version-info.xml" xpointer="v226"/></listitem>
</varlistentry>
<varlistentry>
<term><keycap>q</keycap></term>
<listitem><para>Terminate the program.</para></listitem>
</varlistentry>
<varlistentry>
<term><keycap>p</keycap></term>
<term><keycap>t</keycap></term>
<term><keycap>c</keycap></term>
<term><keycap>m</keycap></term>
<term><keycap>i</keycap></term>
<listitem><para>Sort the control groups by path, number of
tasks, CPU load, memory usage, or I/O load, respectively. This
setting may also be controlled using the
<option>--order=</option> command line
switch.</para></listitem>
</varlistentry>
<varlistentry>
<term><keycap>%</keycap></term>
<listitem><para>Toggle between showing CPU time as time or
percentage. This setting may also be controlled using the
<option>--cpu=</option> command line switch.</para>
<xi:include href="version-info.xml" xpointer="v201"/></listitem>
</varlistentry>
<varlistentry>
<term><keycap>/</keycap></term>
<listitem><para>Toggle the root control group row between
cgroup-sum accounting (default) and whole-system accounting from
<filename>/proc/meminfo</filename>. This setting may also be
controlled using the <option>--root-memory=</option> command line
switch.</para>
<xi:include href="version-info.xml" xpointer="v263"/></listitem>
</varlistentry>
<varlistentry>
<term><keycap>+</keycap></term>
<term><keycap>-</keycap></term>
<listitem><para>Increase or decrease refresh delay,
respectively. This setting may also be controlled using the
<option>--delay=</option> command line
switch.</para></listitem>
</varlistentry>
<varlistentry>
<term><keycap>P</keycap></term>
<listitem><para>Toggle between counting all tasks, or only
userspace processes. This setting may also be controlled using
the <option>-P</option> command line switch (see
above).</para>
<xi:include href="version-info.xml" xpointer="v227"/></listitem>
</varlistentry>
<varlistentry>
<term><keycap>k</keycap></term>
<listitem><para>Toggle between counting all tasks, or only
userspace processes and kernel threads. This setting may also
be controlled using the <option>-k</option> command line
switch (see above).</para>
<xi:include href="version-info.xml" xpointer="v226"/></listitem>
</varlistentry>
<varlistentry>
<term><keycap>r</keycap></term>
<listitem><para>Toggle between recursively including or
excluding processes in child control groups in control group
process counts. This setting may also be controlled using the
<option>--recursive=</option> command line switch. This key is
not available if all tasks are counted, it is only available
if processes are counted, as enabled with the
<keycap>P</keycap> or <keycap>k</keycap>
keys.</para>
<xi:include href="version-info.xml" xpointer="v226"/></listitem>
</varlistentry>
</variablelist>
</refsect1>
<refsect1>
<title>Exit status</title>
<para>On success, 0 is returned, a non-zero failure code
otherwise.</para>
</refsect1>
<refsect1>
<title>See Also</title>
<para><simplelist type="inline">
<member><citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry></member>
<member><citerefentry><refentrytitle>systemctl</refentrytitle><manvolnum>1</manvolnum></citerefentry></member>
<member><citerefentry><refentrytitle>systemd-cgls</refentrytitle><manvolnum>1</manvolnum></citerefentry></member>
<member><citerefentry><refentrytitle>systemd.resource-control</refentrytitle><manvolnum>5</manvolnum></citerefentry></member>
<member><citerefentry project='man-pages'><refentrytitle>top</refentrytitle><manvolnum>1</manvolnum></citerefentry></member>
</simplelist></para>
</refsect1>
</refentry>