code.grnet.gr Git - ganeti-local/blob - docs/hooks.sgml

   1 <!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook V4.2//EN" [
   2 ]>
   3   <article class="specification">
   4   <articleinfo>
   5     <title>Ganeti customisation using hooks</title>
   6   </articleinfo>
   7   <para>Documents ganeti version 1.2</para>
   8   <section>
   9     <title>Introduction</title>
  10
  11     <para>
  12       In order to allow customisation of operations, ganeti will run
  13       scripts under <filename
  14       class="directory">/etc/ganeti/hooks</filename> based on certain
  15       rules.
  16     </para>
  17
  18       <para>This is similar to the <filename
  19       class="directory">/etc/network/</filename> structure present in
  20       Debian for network interface handling.</para>
  21
  22     </section>
  23
  24
  25     <section>
  26       <title>Organisation</title>
  27
  28       <para>For every operation, two sets of scripts are run:
  29
  30       <itemizedlist>
  31           <listitem>
  32             <simpara>pre phase (for authorization/checking)</simpara>
  33           </listitem>
  34           <listitem>
  35             <simpara>post phase (for logging)</simpara>
  36           </listitem>
  37         </itemizedlist>
  38       </para>
  39
  40       <para>Also, for each operation, the scripts are run on one or
  41       more nodes, depending on the operation type.</para>
  42
  43       <para>Note that, even though we call them scripts, we are
  44       actually talking about any executable.</para>
  45
  46       <section>
  47         <title><emphasis>pre</emphasis> scripts</title>
  48
  49         <para>The <emphasis>pre</emphasis> scripts have a definite
  50         target: to check that the operation is allowed given the
  51         site-specific constraints. You could have, for example, a rule
  52         that says every new instance is required to exists in a
  53         database; to implement this, you could write a script that
  54         checks the new instance parameters against your
  55         database.</para>
  56
  57         <para>The objective of these scripts should be their return
  58         code (zero or non-zero for success and failure). However, if
  59         they modify the environment in any way, they should be
  60         idempotent, as failed executions could be restarted and thus
  61         the script(s) run again with exactly the same
  62         parameters.</para>
  63
  64       </section>
  65
  66       <section>
  67         <title><emphasis>post</emphasis> scripts</title>
  68
  69         <para>These scripts should do whatever you need as a reaction
  70         to the completion of an operation. Their return code is not
  71         checked (but logged), and they should not depend on the fact
  72         that the <emphasis>pre</emphasis> scripts have been
  73         run.</para>
  74
  75       </section>
  76
  77       <section>
  78         <title>Naming</title>
  79
  80         <para>The allowed names for the scripts consist of (similar to
  81         <citerefentry> <refentrytitle>run-parts</refentrytitle>
  82         <manvolnum>8</manvolnum> </citerefentry>) upper and lower
  83         case, digits, underscores and hyphens. In other words, the
  84         regexp
  85         <computeroutput>^[a-zA-Z0-9_-]+$</computeroutput>. Also,
  86         non-executable scripts will be ignored.
  87         </para>
  88       </section>
  89
  90       <section>
  91         <title>Order of execution</title>
  92
  93         <para>On a single node, the scripts in a directory are run in
  94         lexicographic order (more exactly, the python string
  95         comparison order). It is advisable to implement the usual
  96         <emphasis>NN-name</emphasis> convention where
  97         <emphasis>NN</emphasis> is a two digit number.</para>
  98
  99         <para>For an operation whose hooks are run on multiple nodes,
 100         there is no specific ordering of nodes with regard to hooks
 101         execution; you should assume that the scripts are run in
 102         parallel on the target nodes (keeping on each node the above
 103         specified ordering).  If you need any kind of inter-node
 104         synchronisation, you have to implement it yourself in the
 105         scripts.</para>
 106
 107       </section>
 108
 109       <section>
 110         <title>Execution environment</title>
 111
 112         <para>The scripts will be run as follows:
 113           <itemizedlist>
 114           <listitem>
 115             <simpara>no command line arguments</simpara>
 116           </listitem>
 117             <listitem>
 118               <simpara>no controlling <acronym>tty</acronym></simpara>
 119             </listitem>
 120             <listitem>
 121               <simpara><varname>stdin</varname> is
 122               actually <filename>/dev/null</filename></simpara>
 123             </listitem>
 124             <listitem>
 125               <simpara><varname>stdout</varname> and
 126               <varname>stderr</varname> are directed to
 127               files</simpara>
 128             </listitem>
 129           <listitem>
 130             <simpara>the <varname>PATH</varname> is reset to
 131             <literal>/sbin:/bin:/usr/sbin:/usr/bin</literal></simpara>
 132           </listitem>
 133           <listitem>
 134             <simpara>the environment is cleared, and only
 135             ganeti-specific variables will be left</simpara>
 136           </listitem>
 137           </itemizedlist>
 138
 139         </para>
 140
 141       <para>All informations about the cluster is passed using
 142       environment variables. Different operations will have sligthly
 143       different environments, but most of the variables are
 144       common.</para>
 145
 146     </section>
 147
 148
 149     <section>
 150       <title>Operation list</title>
 151       <table>
 152         <title>Operation list</title>
 153         <tgroup cols="7">
 154           <colspec>
 155           <colspec>
 156           <colspec>
 157           <colspec>
 158           <colspec>
 159           <colspec colname="prehooks">
 160           <colspec colname="posthooks">
 161           <spanspec namest="prehooks" nameend="posthooks"
 162             spanname="bothhooks">
 163           <thead>
 164             <row>
 165               <entry>Operation ID</entry>
 166               <entry>Directory prefix</entry>
 167               <entry>Description</entry>
 168               <entry>Command</entry>
 169               <entry>Supported env. variables</entry>
 170               <entry><emphasis>pre</emphasis> hooks</entry>
 171               <entry><emphasis>post</emphasis> hooks</entry>
 172             </row>
 173           </thead>
 174           <tbody>
 175             <row>
 176               <entry>OP_INIT_CLUSTER</entry>
 177               <entry><filename class="directory">cluster-init</filename></entry>
 178               <entry>Initialises the cluster</entry>
 179               <entry><computeroutput>gnt-cluster init</computeroutput></entry>
 180               <entry><constant>CLUSTER</constant>, <constant>MASTER</constant></entry>
 181               <entry spanname="bothhooks">master node, cluster name</entry>
 182             </row>
 183             <row>
 184               <entry>OP_MASTER_FAILOVER</entry>
 185               <entry><filename class="directory">master-failover</filename></entry>
 186               <entry>Changes the master</entry>
 187               <entry><computeroutput>gnt-cluster master-failover</computeroutput></entry>
 188               <entry><constant>OLD_MASTER</constant>, <constant>NEW_MASTER</constant></entry>
 189               <entry>the new master</entry>
 190               <entry>all nodes</entry>
 191             </row>
 192             <row>
 193               <entry>OP_ADD_NODE</entry>
 194               <entry><filename class="directory">node-add</filename></entry>
 195               <entry>Adds a new node to the cluster</entry>
 196               <entry><computeroutput>gnt-node add</computeroutput></entry>
 197               <entry><constant>NODE_NAME</constant>, <constant>NODE_PIP</constant>, <constant>NODE_SIP</constant></entry>
 198               <entry>all existing nodes</entry>
 199               <entry>all existing nodes plus the new node</entry>
 200             </row>
 201             <row>
 202               <entry>OP_REMOVE_NODE</entry>
 203               <entry><filename class="directory">node-remove</filename></entry>
 204               <entry>Removes a node from the cluster</entry>
 205               <entry><computeroutput>gnt-node remove</computeroutput></entry>
 206               <entry><constant>NODE_NAME</constant></entry>
 207               <entry spanname="bothhooks">all existing nodes except the removed node</entry>
 208             </row>
 209             <row>
 210               <entry>OP_INSTANCE_ADD</entry>
 211               <entry><filename class="directory">instance-add</filename></entry>
 212               <entry>Creates a new instance</entry>
 213               <entry><computeroutput>gnt-instance add</computeroutput></entry>
 214               <entry><constant>INSTANCE_NAME</constant>, <constant>INSTANCE_PRIMARY</constant>, <constant>INSTANCE_SECONDARIES</constant>, <constant>DISK_TEMPLATE</constant>, <constant>MEM_SIZE</constant>, <constant>DISK_SIZE</constant>, <constant>SWAP_SIZE</constant>, <constant>VCPUS</constant>, <constant>INSTANCE_IP</constant>, <constant>INSTANCE_ADD_MODE</constant>, <constant>SRC_NODE</constant>, <constant>SRC_PATH</constant>, <constant>SRC_IMAGE</constant></entry>
 215               <entry spanname="bothhooks" morerows="4">master node, primary and
 216                    secondary nodes</entry>
 217             </row>
 218             <row>
 219               <entry>OP_BACKUP_EXPORT</entry>
 220               <entry><filename class="directory">instance-export</filename></entry>
 221               <entry>Export the instance</entry>
 222               <entry><computeroutput>gnt-backup export</computeroutput></entry>
 223               <entry><constant>INSTANCE_NAME</constant>, <constant>EXPORT_NODE</constant>, <constant>EXPORT_DO_SHUTDOWN</constant></entry>
 224             </row>
 225             <row>
 226               <entry>OP_INSTANCE_START</entry>
 227               <entry><filename class="directory">instance-start</filename></entry>
 228               <entry>Starts an instance</entry>
 229               <entry><computeroutput>gnt-instance start</computeroutput></entry>
 230               <entry><constant>INSTANCE_NAME</constant>, <constant>INSTANCE_PRIMARY</constant>, <constant>INSTANCE_SECONDARIES</constant>, <constant>FORCE</constant></entry>
 231             </row>
 232             <row>
 233               <entry>OP_INSTANCE_SHUTDOWN</entry>
 234               <entry><filename class="directory">instance-shutdown</filename></entry>
 235               <entry>Stops an instance</entry>
 236               <entry><computeroutput>gnt-instance shutdown</computeroutput></entry>
 237               <entry><constant>INSTANCE_NAME</constant>, <constant>INSTANCE_PRIMARY</constant>, <constant>INSTANCE_SECONDARIES</constant></entry>
 238             </row>
 239             <row>
 240               <entry>OP_INSTANCE_MODIFY</entry>
 241               <entry><filename class="directory">instance-modify</filename></entry>
 242               <entry>Modifies the instance parameters.</entry>
 243               <entry><computeroutput>gnt-instance modify</computeroutput></entry>
 244               <entry><constant>INSTANCE_NAME</constant>, <constant>MEM_SIZE</constant>, <constant>VCPUS</constant>, <constant>INSTANCE_IP</constant></entry>
 245             </row>
 246             <row>
 247               <entry>OP_INSTANCE_FAILOVER</entry>
 248               <entry><filename class="directory">instance-failover</filename></entry>
 249               <entry>Failover an instance</entry>
 250               <entry><computeroutput>gnt-instance start</computeroutput></entry>
 251               <entry><constant>INSTANCE_NAME</constant>, <constant>INSTANCE_PRIMARY</constant>, <constant>INSTANCE_SECONDARIES</constant>, <constant>IGNORE_CONSISTENCY</constant></entry>
 252             </row>
 253             <row>
 254               <entry>OP_INSTANCE_REMOVE</entry>
 255               <entry><filename class="directory">instance-remove</filename></entry>
 256               <entry>Remove an instance</entry>
 257               <entry><computeroutput>gnt-instance remove</computeroutput></entry>
 258               <entry><constant>INSTANCE_NAME</constant>, <constant>INSTANCE_PRIMARY</constant>, <constant>INSTANCE_SECONDARIES</constant></entry>
 259             </row>
 260             <row>
 261               <entry>OP_INSTANCE_ADD_MDDRBD</entry>
 262               <entry><filename class="directory">mirror-add</filename></entry>
 263               <entry>Adds a mirror component</entry>
 264               <entry><computeroutput>gnt-instance add-mirror</computeroutput></entry>
 265               <entry><constant>INSTANCE_NAME</constant>, <constant>NEW_SECONDARY</constant>, <constant>DISK_NAME</constant></entry>
 266             </row>
 267             <row>
 268               <entry>OP_INSTANCE_REMOVE_MDDRBD</entry>
 269               <entry><filename class="directory">mirror-remove</filename></entry>
 270               <entry>Removes a mirror component</entry>
 271               <entry><computeroutput>gnt-instance remove-mirror</computeroutput></entry>
 272               <entry><constant>INSTANCE_NAME</constant>, <constant>OLD_SECONDARY</constant>, <constant>DISK_NAME</constant>, <constant>DISK_ID</constant></entry>
 273             </row>
 274             <row>
 275               <entry>OP_INSTANCE_REPLACE_DISKS</entry>
 276               <entry><filename class="directory">mirror-replace</filename></entry>
 277               <entry>Replace all mirror components</entry>
 278               <entry><computeroutput>gnt-instance replace-disks</computeroutput></entry>
 279               <entry><constant>INSTANCE_NAME</constant>, <constant>OLD_SECONDARY</constant>, <constant>NEW_SECONDARY</constant></entry>
 280
 281             </row>
 282           </tbody>
 283         </tgroup>
 284       </table>
 285     </section>
 286
 287     <section>
 288       <title>Environment variables</title>
 289
 290       <para>Note that all variables listed here are actually prefixed
 291       with <constant>GANETI_</constant> in order to provide a
 292       different namespace.</para>
 293
 294       <section>
 295         <title>Common variables</title>
 296
 297         <para>This is the list of environment variables supported by
 298         all operations:</para>
 299
 300         <variablelist>
 301           <varlistentry>
 302             <term>HOOKS_VERSION</term>
 303             <listitem>
 304               <para>Documents the hooks interface version. In case this
 305             doesnt match what the script expects, it should not
 306             run. The documents conforms to the version
 307             <literal>1</literal>.</para>
 308             </listitem>
 309           </varlistentry>
 310           <varlistentry>
 311             <term>HOOKS_PHASE</term>
 312             <listitem>
 313               <para>one of <constant>PRE</constant> or
 314               <constant>POST</constant> denoting which phase are we
 315               in.</para>
 316             </listitem>
 317           </varlistentry>
 318           <varlistentry>
 319             <term>CLUSTER</term>
 320             <listitem>
 321               <para>the cluster name</para>
 322             </listitem>
 323           </varlistentry>
 324           <varlistentry>
 325             <term>MASTER</term>
 326             <listitem>
 327               <para>the master node</para>
 328             </listitem>
 329           </varlistentry>
 330           <varlistentry>
 331             <term>OP_ID</term>
 332             <listitem>
 333               <para>one of the <constant>OP_*</constant> values from
 334               the table of operations</para>
 335             </listitem>
 336           </varlistentry>
 337           <varlistentry>
 338             <term>OBJECT_TYPE</term>
 339             <listitem>
 340               <para>one of <simplelist type="inline">
 341                   <member><constant>INSTANCE</constant></member>
 342                   <member><constant>NODE</constant></member>
 343                   <member><constant>CLUSTER</constant></member>
 344                 </simplelist>, showing the target of the operation.
 345              </para>
 346             </listitem>
 347           </varlistentry>
 348           <!-- commented out since it causes problems in our rpc
 349                multi-node optimised calls
 350           <varlistentry>
 351             <term>HOST_NAME</term>
 352             <listitem>
 353               <para>The name of the node the hook is run on as known by
 354             the cluster.</para>
 355             </listitem>
 356           </varlistentry>
 357           <varlistentry>
 358             <term>HOST_TYPE</term>
 359             <listitem>
 360               <para>one of <simplelist type="inline">
 361                   <member><constant>MASTER</constant></member>
 362                   <member><constant>NODE</constant></member>
 363                 </simplelist>, showing the role of this node in the cluster.
 364              </para>
 365             </listitem>
 366           </varlistentry>
 367           -->
 368         </variablelist>
 369       </section>
 370
 371       <section>
 372         <title>Specialised variables</title>
 373
 374         <para>This is the list of variables which are specific to one
 375         or more operations.</para>
 376         <variablelist>
 377           <varlistentry>
 378             <term>INSTANCE_NAME</term>
 379             <listitem>
 380               <para>The name of the instance which is the target of
 381               the operation.</para>
 382             </listitem>
 383           </varlistentry>
 384           <varlistentry>
 385             <term>INSTANCE_DISK_TYPE</term>
 386             <listitem>
 387               <para>The disk type for the instance.</para>
 388             </listitem>
 389           </varlistentry>
 390           <varlistentry>
 391             <term>INSTANCE_DISK_SIZE</term>
 392             <listitem>
 393               <para>The (OS) disk size for the instance.</para>
 394             </listitem>
 395           </varlistentry>
 396           <varlistentry>
 397             <term>INSTANCE_OS</term>
 398             <listitem>
 399               <para>The name of the instance OS.</para>
 400             </listitem>
 401           </varlistentry>
 402           <varlistentry>
 403             <term>INSTANCE_PRIMARY</term>
 404             <listitem>
 405               <para>The name of the node which is the primary for the
 406               instance.</para>
 407             </listitem>
 408           </varlistentry>
 409           <varlistentry>
 410             <term>INSTANCE_SECONDARIES</term>
 411             <listitem>
 412               <para>Space-separated list of secondary nodes for the
 413               instance.</para>
 414             </listitem>
 415           </varlistentry>
 416           <varlistentry>
 417             <term>NODE_NAME</term>
 418             <listitem>
 419               <para>The target node of this operation (not the node on
 420               which the hook runs).</para>
 421             </listitem>
 422           </varlistentry>
 423           <varlistentry>
 424             <term>NODE_PIP</term>
 425             <listitem>
 426               <para>The primary IP of the target node (the one over
 427               which inter-node communication is done).</para>
 428             </listitem>
 429           </varlistentry>
 430           <varlistentry>
 431             <term>NODE_SIP</term>
 432             <listitem>
 433               <para>The secondary IP of the target node (the one over
 434               which drbd replication is done). This can be equal to
 435               the primary ip, in case the cluster is not
 436               dual-homed.</para>
 437             </listitem>
 438           </varlistentry>
 439           <varlistentry>
 440             <term>OLD_MASTER</term>
 441             <term>NEW_MASTER</term>
 442             <listitem>
 443               <para>The old, respectively the new master for the
 444               master failover operation.</para>
 445             </listitem>
 446           </varlistentry>
 447           <varlistentry>
 448             <term>FORCE</term>
 449             <listitem>
 450               <para>This is provided by some operations when the user
 451               gave this flag.</para>
 452             </listitem>
 453           </varlistentry>
 454           <varlistentry>
 455             <term>IGNORE_CONSISTENCY</term>
 456             <listitem>
 457               <para>The user has specified this flag. It is used when
 458               failing over instances in case the primary node is
 459               down.</para>
 460             </listitem>
 461           </varlistentry>
 462           <varlistentry>
 463             <term>MEM_SIZE, DISK_SIZE, SWAP_SIZE, VCPUS</term>
 464             <listitem>
 465               <para>The memory, disk, swap size and the number of
 466               processor selected for the instance (in
 467               <command>gnt-instance add</command> or
 468               <command>gnt-instance modify</command>).</para>
 469             </listitem>
 470           </varlistentry>
 471           <varlistentry>
 472             <term>INSTANCE_IP</term>
 473             <listitem>
 474               <para>If defined, the instance IP in the
 475               <command>gnt-instance add</command> and
 476               <command>gnt-instance set</command> commands. If not
 477               defined, it means that no IP has been defined.</para>
 478             </listitem>
 479           </varlistentry>
 480           <varlistentry>
 481             <term>DISK_TEMPLATE</term>
 482             <listitem>
 483               <para>The disk template type when creating the instance.</para>
 484             </listitem>
 485           </varlistentry>
 486           <varlistentry>
 487             <term>INSTANCE_ADD_MODE</term>
 488             <listitem>
 489               <para>The mode of the create: either
 490               <constant>create</constant> for create from scratch or
 491               <constant>import</constant> for restoring from an
 492               exported image.</para>
 493             </listitem>
 494           </varlistentry>
 495           <varlistentry>
 496             <term>SRC_NODE, SRC_PATH, SRC_IMAGE</term>
 497             <listitem>
 498               <para>In case the instance has been added by import,
 499               these variables are defined and point to the source
 500               node, source path (the directory containing the image
 501               and the config file) and the source disk image
 502               file.</para>
 503             </listitem>
 504           </varlistentry>
 505           <varlistentry>
 506             <term>DISK_NAME</term>
 507             <listitem>
 508               <para>The disk name (either <filename>sda</filename> or
 509               <filename>sdb</filename>) in mirror operations
 510               (add/remove mirror).</para>
 511             </listitem>
 512           </varlistentry>
 513           <varlistentry>
 514             <term>DISK_ID</term>
 515             <listitem>
 516               <para>The disk id for mirror remove operations. You can
 517               look this up using <command>gnt-instance
 518               info</command>.</para>
 519             </listitem>
 520           </varlistentry>
 521           <varlistentry>
 522             <term>NEW_SECONDARY</term>
 523             <listitem>
 524               <para>The name of the node on which the new mirror
 525               componet is being added. This can be the name of the
 526               current secondary, if the new mirror is on the same
 527               secondary.</para>
 528             </listitem>
 529           </varlistentry>
 530           <varlistentry>
 531             <term>OLD_SECONDARY</term>
 532             <listitem>
 533               <para>The name of the old secondary. This is used in
 534               both <command>replace-disks</command> and
 535               <command>remove-mirror</command>. Note that this can be
 536               equal to the new secondary (only
 537               <command>replace-disks</command> has both variables) if
 538               the secondary node hasn't actually changed).</para>
 539             </listitem>
 540           </varlistentry>
 541           <varlistentry>
 542             <term>EXPORT_NODE</term>
 543             <listitem>
 544               <para>The node on which the exported image of the
 545               instance was done.</para>
 546             </listitem>
 547           </varlistentry>
 548           <varlistentry>
 549             <term>EXPORT_DO_SHUTDOWN</term>
 550             <listitem>
 551               <para>This variable tells if the instance has been
 552               shutdown or not while doing the export. In the "was
 553               shutdown" case, it's likely that the filesystem is
 554               consistent, whereas in the "did not shutdown" case, the
 555               filesystem would need a check (journal replay or full
 556               fsck) in order to guarantee consistency.</para>
 557             </listitem>
 558           </varlistentry>
 559         </variablelist>
 560
 561       </section>
 562
 563     </section>
 564
 565   </section>
 566   </article>