/usr/share/doc/ganeti/html/design-partitioned.html is in ganeti-doc 2.16.0~rc2-1build1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 | <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>Partitioned Ganeti — Ganeti 2.16.0~rc2 documentation</title>
<link rel="stylesheet" href="_static/style.css" type="text/css" />
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
<script type="text/javascript">
var DOCUMENTATION_OPTIONS = {
URL_ROOT: './',
VERSION: '2.16.0~rc2',
COLLAPSE_INDEX: false,
FILE_SUFFIX: '.html',
HAS_SOURCE: true,
SOURCELINK_SUFFIX: '.txt'
};
</script>
<script type="text/javascript" src="_static/jquery.js"></script>
<script type="text/javascript" src="_static/underscore.js"></script>
<script type="text/javascript" src="_static/doctools.js"></script>
<link rel="search" title="Search" href="search.html" />
<link rel="next" title="Performance tests for QA" href="design-performance-tests.html" />
<link rel="prev" title="Ganeti Instance Import/Export using Open Virtualization Format" href="design-ovf-support.html" />
</head>
<body>
<div class="related" role="navigation" aria-label="related navigation">
<h3>Navigation</h3>
<ul>
<li class="right" style="margin-right: 10px">
<a href="design-performance-tests.html" title="Performance tests for QA"
accesskey="N">next</a></li>
<li class="right" >
<a href="design-ovf-support.html" title="Ganeti Instance Import/Export using Open Virtualization Format"
accesskey="P">previous</a> |</li>
<li class="nav-item nav-item-0"><a href="index.html">Ganeti 2.16.0~rc2 documentation</a> »</li>
</ul>
</div>
<div class="document">
<div class="documentwrapper">
<div class="bodywrapper">
<div class="body" role="main">
<div class="section" id="partitioned-ganeti">
<h1><a class="toc-backref" href="#id1">Partitioned Ganeti</a><a class="headerlink" href="#partitioned-ganeti" title="Permalink to this headline">¶</a></h1>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name" />
<col class="field-body" />
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Created:</th><td class="field-body">2012-Oct-05</td>
</tr>
<tr class="field-even field"><th class="field-name">Status:</th><td class="field-body">Implemented</td>
</tr>
<tr class="field-odd field"><th class="field-name">Ganeti-Version:</th><td class="field-body">2.7.0, 2.8.0, 2.9.0</td>
</tr>
</tbody>
</table>
<div class="contents topic" id="contents">
<p class="topic-title first">Contents</p>
<ul class="simple">
<li><a class="reference internal" href="#partitioned-ganeti" id="id1">Partitioned Ganeti</a><ul>
<li><a class="reference internal" href="#current-state-and-shortcomings" id="id2">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#proposed-changes" id="id3">Proposed changes</a><ul>
<li><a class="reference internal" href="#exclusive-use-of-disks" id="id4">Exclusive use of disks</a><ul>
<li><a class="reference internal" href="#spindles-as-a-resource" id="id5">Spindles as a resource</a></li>
</ul>
</li>
<li><a class="reference internal" href="#dedicated-cpus" id="id6">Dedicated CPUs</a></li>
<li><a class="reference internal" href="#dedicated-ram" id="id7">Dedicated RAM</a><ul>
<li><a class="reference internal" href="#xen" id="id8">Xen</a></li>
<li><a class="reference internal" href="#kvm" id="id9">KVM</a></li>
<li><a class="reference internal" href="#numa" id="id10">NUMA</a></li>
</ul>
</li>
<li><a class="reference internal" href="#constrained-instance-sizes" id="id11">Constrained instance sizes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#implementation-order" id="id12">Implementation order</a></li>
<li><a class="reference internal" href="#possible-future-enhancements" id="id13">Possible future enhancements</a><ul>
<li><a class="reference internal" href="#network-bandwidth" id="id14">Network bandwidth</a></li>
<li><a class="reference internal" href="#cpu-pinning" id="id15">CPU pinning</a></li>
<li><a class="reference internal" href="#overcommit-for-ram-and-disks" id="id16">Overcommit for RAM and disks</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</div>
<div class="section" id="current-state-and-shortcomings">
<h2><a class="toc-backref" href="#id2">Current state and shortcomings</a><a class="headerlink" href="#current-state-and-shortcomings" title="Permalink to this headline">¶</a></h2>
<p>Currently Ganeti can be used to easily share a node between multiple
virtual instances. While it’s easy to do a completely “best effort”
sharing it’s quite harder to completely reserve resources for the use of
a particular instance. In particular this has to be done manually for
CPUs and disk, is implemented for RAM under Xen, but not under KVM, and
there’s no provision for network level QoS.</p>
</div>
<div class="section" id="proposed-changes">
<h2><a class="toc-backref" href="#id3">Proposed changes</a><a class="headerlink" href="#proposed-changes" title="Permalink to this headline">¶</a></h2>
<p>We want to make it easy to partition a node between machines with
exclusive use of hardware resources. While some sharing will anyway need
to happen (e.g. for operations that use the host domain, or use
resources, like buses, which are unique or very scarce on host systems)
we’ll strive to maintain contention at a minimum, but won’t try to avoid
all possible sources of it.</p>
<div class="section" id="exclusive-use-of-disks">
<h3><a class="toc-backref" href="#id4">Exclusive use of disks</a><a class="headerlink" href="#exclusive-use-of-disks" title="Permalink to this headline">¶</a></h3>
<p><code class="docutils literal"><span class="pre">exclusive_storage</span></code> is a new node parameter. When it’s enabled, Ganeti
will allocate entire disks to instances. Though it’s possible to think
of ways of doing something similar for other storage back-ends, this
design targets only <code class="docutils literal"><span class="pre">plain</span></code> and <code class="docutils literal"><span class="pre">drbd</span></code>. The name is generic enough
in case the feature will be extended to other back-ends. The flag value
should be homogeneous within a node-group; <code class="docutils literal"><span class="pre">cluster-verify</span></code> will report
any violation of this condition.</p>
<p>Ganeti will consider each physical volume in the destination volume
group as a host disk (for proper isolation, an administrator should
make sure that there aren’t multiple PVs on the same physical
disk). When <code class="docutils literal"><span class="pre">exclusive_storage</span></code> is enabled in a node group, all PVs
in the node group must have the same size (within a certain margin, say
1%, defined through a new parameter). Ganeti will check this condition
when the <code class="docutils literal"><span class="pre">exclusive_storage</span></code> flag is set, whenever a new node is added
and as part of <code class="docutils literal"><span class="pre">cluster-verify</span></code>.</p>
<p>When creating a new disk for an instance, Ganeti will allocate the
minimum number of PVs to hold the disk, and those PVs will be excluded
from the pool of available PVs for further disk creations. The
underlying LV will be striped, when striping is allowed by the current
configuration. Ganeti will continue to track only the LVs, and query the
LVM layer to figure out which PVs are available and how much space is
free. Yet, creation, disk growing, and free-space reporting will ignore
any partially allocated PVs, so that PVs won’t be shared between
instance disks.</p>
<p>For compatibility with the DRBD template and to take into account disk
variability, Ganeti will always subtract 2% (this will be a parameter)
from the PV space when calculating how many PVs are needed to allocate
an instance and when nodes report free space.</p>
<p>The obvious target for this option is plain disk template, which doesn’t
provide redundancy. An administrator can still provide resilience
against disk failures by setting up RAID under PVs, but this is
transparent to Ganeti.</p>
<div class="section" id="spindles-as-a-resource">
<h4><a class="toc-backref" href="#id5">Spindles as a resource</a><a class="headerlink" href="#spindles-as-a-resource" title="Permalink to this headline">¶</a></h4>
<p>When resources are dedicated and there are more spindles than instances
on a node, it is natural to assign more spindles to instances than what
is strictly needed. For this reason, we introduce a new resource:
spindles. A spindle is a PV in LVM. The number of spindles required for
a disk of an instance is specified together with the size. Specifying
the number of spindles is possible only when <code class="docutils literal"><span class="pre">exclusive_storage</span></code> is
enabled. It is an error to specify a number of spindles insufficient to
contain the requested disk size.</p>
<p>When <code class="docutils literal"><span class="pre">exclusive_storage</span></code> is not enabled, spindles are not used in free
space calculation, in allocation algorithms, and policies. When it’s
enabled, <code class="docutils literal"><span class="pre">hspace</span></code>, <code class="docutils literal"><span class="pre">hbal</span></code>, and allocators will use spindles instead
of disk size for their computation. For each node, the number of all the
spindles in every LVM group is recorded, and different LVM groups are
accounted separately in allocation and balancing.</p>
<p>There is already a concept of spindles in Ganeti. It’s not related to
any actual spindle or volume count, but it’s used in <code class="docutils literal"><span class="pre">spindle_use</span></code> to
measure the pressure of an instance on the storage system and in
<code class="docutils literal"><span class="pre">spindle_ratio</span></code> to balance the I/O load on the nodes. When
<code class="docutils literal"><span class="pre">exclusive_storage</span></code> is enabled, these parameters as currently defined
won’t make any sense, so their meaning will be changed in this way:</p>
<ul class="simple">
<li><code class="docutils literal"><span class="pre">spindle_use</span></code> refers to the resource, hence to the actual spindles
(PVs in LVM), used by an instance. The values specified in the instance
policy specifications are compared to the run-time numbers of spindle
used by an instance. The <code class="docutils literal"><span class="pre">spindle_use</span></code> back-end parameter will be
ignored.</li>
<li><code class="docutils literal"><span class="pre">spindle_ratio</span></code> in instance policies and <code class="docutils literal"><span class="pre">spindle_count</span></code> in node
parameters are ignored, as the exclusive assignment of PVs already
implies a value of 1.0 for the first, and the second is replaced by
the actual number of spindles.</li>
</ul>
<p>When <code class="docutils literal"><span class="pre">exclusive_storage</span></code> is disabled, the existing spindle parameters
behave as before.</p>
</div>
</div>
<div class="section" id="dedicated-cpus">
<h3><a class="toc-backref" href="#id6">Dedicated CPUs</a><a class="headerlink" href="#dedicated-cpus" title="Permalink to this headline">¶</a></h3>
<p><code class="docutils literal"><span class="pre">vpcu_ratio</span></code> can be used to tie the number of VCPUs to the number of
CPUs provided by the hardware. We need to take into account the CPU
usage of the hypervisor. For Xen, this means counting the number of
VCPUs assigned to <code class="docutils literal"><span class="pre">Domain-0</span></code>.</p>
<p>For KVM, it’s more difficult to limit the number of CPUs used by the
node OS. <code class="docutils literal"><span class="pre">cgroups</span></code> could be a solution to restrict the node OS to use
some of the CPUs, leaving the other ones to instances and KVM processes.
For KVM, the number of CPUs for the host system should also be a
hypervisor parameter (set at the node group level).</p>
</div>
<div class="section" id="dedicated-ram">
<h3><a class="toc-backref" href="#id7">Dedicated RAM</a><a class="headerlink" href="#dedicated-ram" title="Permalink to this headline">¶</a></h3>
<p>Instances should not compete for RAM. This is easily done on Xen, but it
is tricky on KVM.</p>
<div class="section" id="xen">
<h4><a class="toc-backref" href="#id8">Xen</a><a class="headerlink" href="#xen" title="Permalink to this headline">¶</a></h4>
<p>Memory is already fully segregated under Xen, if sharing mechanisms
(transcendent memory, auto ballooning, etc) are not in use.</p>
</div>
<div class="section" id="kvm">
<h4><a class="toc-backref" href="#id9">KVM</a><a class="headerlink" href="#kvm" title="Permalink to this headline">¶</a></h4>
<p>Under KVM or LXC memory is fully shared between the host system and all
the guests, and instances can even be swapped out by the host OS.</p>
<p>It’s not clear if the problem can be solved by limiting the size of the
instances, so that there is plenty of room for the host OS.</p>
<p>We could implement segregation using cgroups to limit the memory used by
the host OS. This requires finishing the implementation of the memory
hypervisor status (set at the node group level) that changes how free
memory is computed under KVM systems. Then we have to add a way to
enforce this limit on the host system itself, rather than leaving it as
a calculation tool only.</p>
<p>Another problem for KVM is that we need to decide about the size of the
cgroup versus the size of the VM: some overhead will in particular
exist, due to the fact that an instance and its encapsulating KVM
process share the same space. For KVM systems the physical memory
allocatable to instances should be computed by subtracting an overhead
for the KVM processes, whose value can be either statically configured
or set in a hypervisor status parameter.</p>
</div>
<div class="section" id="numa">
<h4><a class="toc-backref" href="#id10">NUMA</a><a class="headerlink" href="#numa" title="Permalink to this headline">¶</a></h4>
<p>If instances are pinned to CPUs, and the amount of memory used for every
instance is proportionate to the number of VCPUs, NUMA shouldn’t be a
problem, as the hypervisors allocate memory in the appropriate NUMA
node. Work is in progress in Xen and the Linux kernel to always allocate
memory correctly even without pinning. Therefore, we don’t need to
address this problem specifically; it will be solved by future versions
of the hypervisors or by implementing CPU pinning.</p>
</div>
</div>
<div class="section" id="constrained-instance-sizes">
<h3><a class="toc-backref" href="#id11">Constrained instance sizes</a><a class="headerlink" href="#constrained-instance-sizes" title="Permalink to this headline">¶</a></h3>
<p>In order to simplify allocation and resource provisioning we want to
limit the possible sizes of instances to a finite set of specifications,
defined at node-group level.</p>
<p>Currently it’s possible to define an instance policy that limits the
minimum and maximum value for CPU, memory, and disk usage (and spindles
and any other resource, when implemented), independently from each other. We
extend the policy by allowing it to contain more occurrences of the
specifications for both the limits for the instance resources. Each
specification pair (minimum and maximum) has a unique priority
associated to it (or in other words, specifications are ordered), which
is used by <code class="docutils literal"><span class="pre">hspace</span></code> (see below). The standard specification doesn’t
change: there is one for the whole cluster.</p>
<p>For example, a policy could be set up to allow instances with this
constraints:</p>
<ul class="simple">
<li>between 1 and 2 CPUs, 2 GB of RAM, and between 10 GB and 400 GB of
disk space;</li>
<li>4 CPUs, 4 GB of RAM, and between 10 GB and 800 GB of disk space.</li>
</ul>
<p>Then, an instance using 1 CPU, 2 GB of RAM and 50 GB of disk would be
legal, as an instance using 4 CPUs, 4 GB of RAM, and 20 GB of disk,
while an instance using 2 CPUs, 4 GB of RAM and 40 GB of disk would be
illegal.</p>
<p>Ganeti will refuse to create (or modify) instances that violate instance
policy constraints, unless the flag <code class="docutils literal"><span class="pre">--ignore-ipolicy</span></code> is passed.</p>
<p>While the changes needed to check constraint violations are
straightforward, <code class="docutils literal"><span class="pre">hspace</span></code> behavior needs some adjustments for tiered
allocation. <code class="docutils literal"><span class="pre">hspace</span></code> will start to allocate instances using the
maximum specification with the highest priority, then it will try to
lower the most constrained resources (without breaking the policy)
before moving to the second highest priority, and so on.</p>
<p>For consistent results in capacity calculation, the specifications
inside a policy should be ordered so that the biggest specifications
have the highest priorities. Also, specifications should not overlap.
Ganeti won’t check nor enforce such constraints, though.</p>
</div>
</div>
<div class="section" id="implementation-order">
<h2><a class="toc-backref" href="#id12">Implementation order</a><a class="headerlink" href="#implementation-order" title="Permalink to this headline">¶</a></h2>
<p>We will implement this design in the following order:</p>
<ul class="simple">
<li>Exclusive use of disks (without spindles as a resource)</li>
<li>Constrained instance sizes</li>
<li>Spindles as a resource</li>
<li>Dedicated CPU and memory</li>
</ul>
<p>In this way have always new features that are immediately useful.
Spindles as a resource are not needed for correct capacity calculation,
as long as allowed disk sizes are multiples of spindle size, so it’s
been moved after constrained instance sizes. If it turns out that it’s
easier to implement dedicated disks with spindles as a resource, then we
will do that.</p>
</div>
<div class="section" id="possible-future-enhancements">
<h2><a class="toc-backref" href="#id13">Possible future enhancements</a><a class="headerlink" href="#possible-future-enhancements" title="Permalink to this headline">¶</a></h2>
<p>This section briefly describes some enhancements to the current design.
They may require their own design document, and must be re-evaluated
when considered for implementation, as Ganeti and the hypervisors may
change substantially in the meantime.</p>
<div class="section" id="network-bandwidth">
<h3><a class="toc-backref" href="#id14">Network bandwidth</a><a class="headerlink" href="#network-bandwidth" title="Permalink to this headline">¶</a></h3>
<p>A new resource is introduced: network bandwidth. An administrator must
be able to assign some network bandwidth to the virtual interfaces of an
instance, and set limits in instance policies. Also, a list of the
physical network interfaces available for Ganeti use and their maximum
bandwidth must be kept at node-group or node level. This information
will be taken into account for allocation, balancing, and free-space
calculation.</p>
<p>An additional enhancement is Ganeti enforcing the values set in the
bandwidth resource. This can be done by configuring limits for example
via openvswitch or normal QoS for bridging or routing. The bandwidth
resource represents the average bandwidth usage, so a few new back-end
parameters are needed to configure how to deal with bursts (they depend
on the actual way used to enforce the limit).</p>
</div>
<div class="section" id="cpu-pinning">
<h3><a class="toc-backref" href="#id15">CPU pinning</a><a class="headerlink" href="#cpu-pinning" title="Permalink to this headline">¶</a></h3>
<p>In order to avoid unwarranted migrations between CPUs and to deal with
NUMA effectively we may need CPU pinning. CPU scheduling is a complex
topic and still under active development in Xen and the Linux kernel, so
we wont’ try to outsmart their developers. If we need pinning it’s more
to have predictable performance than to get the maximum performance
(which is best done by the hypervisor), so we’ll implement a very simple
algorithm that allocates CPUs when an instance is assigned to a node
(either when it’s created or when it’s moved) and takes into account
NUMA and maybe CPU multithreading. A more refined version might run also
when an instance is deleted, but that would involve reassigning CPUs,
which could be bad with NUMA.</p>
</div>
<div class="section" id="overcommit-for-ram-and-disks">
<h3><a class="toc-backref" href="#id16">Overcommit for RAM and disks</a><a class="headerlink" href="#overcommit-for-ram-and-disks" title="Permalink to this headline">¶</a></h3>
<p>Right now it is possible to assign more VCPUs to the instances running
on a node than there are CPU available. This works as normally CPU usage
on average is way below 100%. There are ways to share memory pages
(e.g. KSM, transcendent memory) and disk blocks, so we could add new
parameters to overcommit memory and disks, similar to <code class="docutils literal"><span class="pre">vcpu_ratio</span></code>.</p>
</div>
</div>
</div>
</div>
</div>
</div>
<div class="sphinxsidebar" role="navigation" aria-label="main navigation">
<div class="sphinxsidebarwrapper">
<h3><a href="index.html">Table Of Contents</a></h3>
<ul>
<li><a class="reference internal" href="#">Partitioned Ganeti</a><ul>
<li><a class="reference internal" href="#current-state-and-shortcomings">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#proposed-changes">Proposed changes</a><ul>
<li><a class="reference internal" href="#exclusive-use-of-disks">Exclusive use of disks</a><ul>
<li><a class="reference internal" href="#spindles-as-a-resource">Spindles as a resource</a></li>
</ul>
</li>
<li><a class="reference internal" href="#dedicated-cpus">Dedicated CPUs</a></li>
<li><a class="reference internal" href="#dedicated-ram">Dedicated RAM</a><ul>
<li><a class="reference internal" href="#xen">Xen</a></li>
<li><a class="reference internal" href="#kvm">KVM</a></li>
<li><a class="reference internal" href="#numa">NUMA</a></li>
</ul>
</li>
<li><a class="reference internal" href="#constrained-instance-sizes">Constrained instance sizes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#implementation-order">Implementation order</a></li>
<li><a class="reference internal" href="#possible-future-enhancements">Possible future enhancements</a><ul>
<li><a class="reference internal" href="#network-bandwidth">Network bandwidth</a></li>
<li><a class="reference internal" href="#cpu-pinning">CPU pinning</a></li>
<li><a class="reference internal" href="#overcommit-for-ram-and-disks">Overcommit for RAM and disks</a></li>
</ul>
</li>
</ul>
</li>
</ul>
<h4>Previous topic</h4>
<p class="topless"><a href="design-ovf-support.html"
title="previous chapter">Ganeti Instance Import/Export using Open Virtualization Format</a></p>
<h4>Next topic</h4>
<p class="topless"><a href="design-performance-tests.html"
title="next chapter">Performance tests for QA</a></p>
<div role="note" aria-label="source link">
<h3>This Page</h3>
<ul class="this-page-menu">
<li><a href="_sources/design-partitioned.rst.txt"
rel="nofollow">Show Source</a></li>
</ul>
</div>
<div id="searchbox" style="display: none" role="search">
<h3>Quick search</h3>
<form class="search" action="search.html" method="get">
<div><input type="text" name="q" /></div>
<div><input type="submit" value="Go" /></div>
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
<script type="text/javascript">$('#searchbox').show(0);</script>
</div>
</div>
<div class="clearer"></div>
</div>
<div class="related" role="navigation" aria-label="related navigation">
<h3>Navigation</h3>
<ul>
<li class="right" style="margin-right: 10px">
<a href="design-performance-tests.html" title="Performance tests for QA"
>next</a></li>
<li class="right" >
<a href="design-ovf-support.html" title="Ganeti Instance Import/Export using Open Virtualization Format"
>previous</a> |</li>
<li class="nav-item nav-item-0"><a href="index.html">Ganeti 2.16.0~rc2 documentation</a> »</li>
</ul>
</div>
<div class="footer" role="contentinfo">
© Copyright 2018, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015 Google Inc..
Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.6.7.
</div>
</body>
</html>
|