Skip to content

Commit

Permalink
Add 'Reserving Compute Resources'
Browse files Browse the repository at this point in the history
* Add overview of reserving compute resources. The actual
  details are described well in the Velum dashboard, therefore
  no need to go into the details too much.
  • Loading branch information
svenseeberg committed Oct 29, 2018
1 parent b82a0dc commit c8b00a0
Show file tree
Hide file tree
Showing 2 changed files with 49 additions and 0 deletions.
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
49 changes: 49 additions & 0 deletions xml/admin_administration.xml
Expand Up @@ -1304,4 +1304,53 @@ worker-4 Ready,SchedulingDisabled <none> 21h v1.9.8
</procedure>
</sect2>
</sect1>

<sect1 xml:id="sec.admin.compute_resources">
<title>Reserving Compute Resources</title>
<para>
By default, &kube; will allocate all available hardware resources of
a node to pods. This can starve core services of needed resources,
which are, for example, required for managing single nodes or the
cluster. To prevent core services from running out of resources, you
can reserve CPU, memory, and disk resources for them.
</para>
<warning>
<title>Carefully Check Entered Values</title>
<para>
Entering invalid values into the input fields may break nodes.
Carefully check the entered values before selecting the
<guimenu>Save</guimenu> button.
</para>
</warning>
<informalfigure>
<mediaobject>
<imageobject role="fo">
<imagedata fileref="velum_settings_compute_resource.png" width="100%"/>
</imageobject>
<imageobject role="html">
<imagedata fileref="velum_settings_compute_resource.png" width="100%"/>
</imageobject>
</mediaobject>
</informalfigure>
<para>
To reserve hardware resources, go to the &dashboard; dashboard and
then proceed to <emphasis>Settings</emphasis> and <emphasis>Compute
Resources Reservation</emphasis>.
</para>
<para>
You can reserve resources for &kube; services in the box
<literal>&kube; core services</literal> and for services running on
a single node in <literal>Host system services</literal>.
</para>
<para>
In the box <literal>Eviction threshold</literal>, you can set rules
for killing pods when the usage of RAM or storage reaches a defined
level. This prevents nodes from actually running out of resources,
which would then trigger the default out-of-resource-handling.
</para>
<para>
To activate entered settings, use the <guimenu>Save</guimenu> button
at the bottom of the page.
</para>
</sect1>
</chapter>

0 comments on commit c8b00a0

Please sign in to comment.