.. _resource-limits:

Resource limits
###############

Linux systems employ limiting or quota mechanisms to provide quality of
service for system resources and contain rogue processes.

These limits are layered at the system-level and user-level. If these limits
need to be modified, it is useful to understand the different limit
configurations.

.. contents:: :local:
    :depth: 2


System-wide limits
==================

Some global resource limits are implemented in the Linux kernel and are
controllable with kernel parameters.

For example, a global limit for the maximum number of open files is set with
the *fs.file-max* parameter. This limit applies to all processes and users an
cannot be exceeded other limit values.

Checking limit
**************

You can check a current value with :command:`sysctl -n <PARAMETER>`. For
example:

.. code:: bash

   sysctl -n fs.file-max


This *fs.file-max* value is set intentionally high on |CL| systems by
default. You can check the maximum value supported by the system with:

.. code::

   cat /proc/sys/fs/file-max


Overriding limit
****************

You can override a value with :command:`sysctl -w <PARAMETER>`. For
example:

.. code:: bash

   sudo sysctl -w fs.file-max=<NUMBER>

If needed permanently, the value can be set by creating a
:file:`/etc/sysctl.d/*.conf` file (see :command:`man sysctl.d` for details).
For example:

.. code:: bash

   sudo mkdir -p /etc/sysctl.d/

   sudo tee /etc/sysctl.d/fs-file-max.conf  > /dev/null <<'EOF'
   fs.file-max=<NUMBER>
   EOF





Per-user limits
===============

For processes not managed by systemd, resource limits can be set for PAM
logins on a per-user basis with upper and lower limits in the
:file:`/etc/security/limits.conf` file.

You can set temporary values and check the current values with the
:command:`ulimit` command. For example, to change the soft limit of maximum
number of open file descriptors for the current user:

.. code::

   ulimit -S -n <NUMBER>

See :command:`man limits.conf` for details.


Service limits
==============

Resource limits for services started with systemd units do not follow normal
user limits because the process is started in a separate `Linux control group
(cgroup) <https://www.kernel.org/doc/Documentation/cgroup-v2.txt>`_ Linux
cgroups associate related process groups and provide resource accounting.

Resource limits for individual systemd services can be controlled inside their
unit files or its configuration drop-in directory with the resource Limit
directives. See `process properties section of the systemd.exec man page
<https://www.freedesktop.org/software/systemd/man/systemd.exec.html>`_.

Resource limits for all systemd services can be controlled with a file in the
:file:`/etc/systemd/system.conf.d/` directory. For example, to have no
restriction on the number of open files:

.. code::

   sudo mkdir -p /etc/systemd/system.conf.d/

   sudo tee /etc/systemd/system.conf.d/50-nfiles.conf  > /dev/null <<'EOF'
   [Manager]
   DefaultLimitNOFILE=infinity
   EOF