1 <!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook V4.2//EN" [
3 <article class="specification">
5 <title>Ganeti customisation using hooks</title>
7 <para>Documents ganeti version 1.2</para>
9 <title>Introduction</title>
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
18 <para>This is similar to the <filename
19 class="directory">/etc/network/</filename> structure present in
20 Debian for network interface handling.</para>
25 <title>Organisation</title>
27 <para>For every operation, two sets of scripts are run:
31 <simpara>pre phase (for authorization/checking)</simpara>
34 <simpara>post phase (for logging)</simpara>
39 <para>Also, for each operation, the scripts are run on one or
40 more nodes, depending on the operation type.</para>
42 <para>Note that, even though we call them scripts, we are
43 actually talking about any executable.</para>
46 <title><emphasis>pre</emphasis> scripts</title>
48 <para>The <emphasis>pre</emphasis> scripts have a definite
49 target: to check that the operation is allowed given the
50 site-specific constraints. You could have, for example, a rule
51 that says every new instance is required to exists in a
52 database; to implement this, you could write a script that
53 checks the new instance parameters against your
56 <para>The objective of these scripts should be their return
57 code (zero or non-zero for success and failure). However, if
58 they modify the environment in any way, they should be
59 idempotent, as failed executions could be restarted and thus
60 the script(s) run again with exactly the same
64 Note that if a node is unreachable at the time a hooks is run,
65 this will not be interpreted as a deny for the execution. In
66 other words, only an actual error returned from a script will
67 cause abort, and not an unreachable node.
71 Therefore, if you want to guarantee that a hook script is run
72 and denies an action, it's best to put it on the master node.
78 <title><emphasis>post</emphasis> scripts</title>
80 <para>These scripts should do whatever you need as a reaction
81 to the completion of an operation. Their return code is not
82 checked (but logged), and they should not depend on the fact
83 that the <emphasis>pre</emphasis> scripts have been
91 <para>The allowed names for the scripts consist of (similar to
92 <citerefentry> <refentrytitle>run-parts</refentrytitle>
93 <manvolnum>8</manvolnum> </citerefentry>) upper and lower
94 case, digits, underscores and hyphens. In other words, the
96 <computeroutput>^[a-zA-Z0-9_-]+$</computeroutput>. Also,
97 non-executable scripts will be ignored.
102 <title>Order of execution</title>
104 <para>On a single node, the scripts in a directory are run in
105 lexicographic order (more exactly, the python string
106 comparison order). It is advisable to implement the usual
107 <emphasis>NN-name</emphasis> convention where
108 <emphasis>NN</emphasis> is a two digit number.</para>
110 <para>For an operation whose hooks are run on multiple nodes,
111 there is no specific ordering of nodes with regard to hooks
112 execution; you should assume that the scripts are run in
113 parallel on the target nodes (keeping on each node the above
114 specified ordering). If you need any kind of inter-node
115 synchronisation, you have to implement it yourself in the
121 <title>Execution environment</title>
123 <para>The scripts will be run as follows:
126 <simpara>no command line arguments</simpara>
129 <simpara>no controlling <acronym>tty</acronym></simpara>
132 <simpara><varname>stdin</varname> is
133 actually <filename>/dev/null</filename></simpara>
136 <simpara><varname>stdout</varname> and
137 <varname>stderr</varname> are directed to
141 <simpara>the <varname>PATH</varname> is reset to
142 <literal>/sbin:/bin:/usr/sbin:/usr/bin</literal></simpara>
145 <simpara>the environment is cleared, and only
146 ganeti-specific variables will be left</simpara>
152 <para>All informations about the cluster is passed using
153 environment variables. Different operations will have sligthly
154 different environments, but most of the variables are
161 <title>Operation list</title>
163 <title>Operation list</title>
170 <colspec colname="prehooks">
171 <colspec colname="posthooks">
172 <spanspec namest="prehooks" nameend="posthooks"
173 spanname="bothhooks">
176 <entry>Operation ID</entry>
177 <entry>Directory prefix</entry>
178 <entry>Description</entry>
179 <entry>Command</entry>
180 <entry>Supported env. variables</entry>
181 <entry><emphasis>pre</emphasis> hooks</entry>
182 <entry><emphasis>post</emphasis> hooks</entry>
187 <entry>OP_INIT_CLUSTER</entry>
188 <entry><filename class="directory">cluster-init</filename></entry>
189 <entry>Initialises the cluster</entry>
190 <entry><computeroutput>gnt-cluster init</computeroutput></entry>
191 <entry><constant>CLUSTER</constant>, <constant>MASTER</constant></entry>
192 <entry spanname="bothhooks">master node, cluster name</entry>
195 <entry>OP_MASTER_FAILOVER</entry>
196 <entry><filename class="directory">master-failover</filename></entry>
197 <entry>Changes the master</entry>
198 <entry><computeroutput>gnt-cluster master-failover</computeroutput></entry>
199 <entry><constant>OLD_MASTER</constant>, <constant>NEW_MASTER</constant></entry>
200 <entry>the new master</entry>
201 <entry>all nodes</entry>
204 <entry>OP_ADD_NODE</entry>
205 <entry><filename class="directory">node-add</filename></entry>
206 <entry>Adds a new node to the cluster</entry>
207 <entry><computeroutput>gnt-node add</computeroutput></entry>
208 <entry><constant>NODE_NAME</constant>, <constant>NODE_PIP</constant>, <constant>NODE_SIP</constant></entry>
209 <entry>all existing nodes</entry>
210 <entry>all existing nodes plus the new node</entry>
213 <entry>OP_REMOVE_NODE</entry>
214 <entry><filename class="directory">node-remove</filename></entry>
215 <entry>Removes a node from the cluster</entry>
216 <entry><computeroutput>gnt-node remove</computeroutput></entry>
217 <entry><constant>NODE_NAME</constant></entry>
218 <entry spanname="bothhooks">all existing nodes except the removed node</entry>
221 <entry>OP_INSTANCE_ADD</entry>
222 <entry><filename class="directory">instance-add</filename></entry>
223 <entry>Creates a new instance</entry>
224 <entry><computeroutput>gnt-instance add</computeroutput></entry>
225 <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>
226 <entry spanname="bothhooks" morerows="4">master node, primary and
227 secondary nodes</entry>
230 <entry>OP_BACKUP_EXPORT</entry>
231 <entry><filename class="directory">instance-export</filename></entry>
232 <entry>Export the instance</entry>
233 <entry><computeroutput>gnt-backup export</computeroutput></entry>
234 <entry><constant>INSTANCE_NAME</constant>, <constant>EXPORT_NODE</constant>, <constant>EXPORT_DO_SHUTDOWN</constant></entry>
237 <entry>OP_INSTANCE_START</entry>
238 <entry><filename class="directory">instance-start</filename></entry>
239 <entry>Starts an instance</entry>
240 <entry><computeroutput>gnt-instance start</computeroutput></entry>
241 <entry><constant>INSTANCE_NAME</constant>, <constant>INSTANCE_PRIMARY</constant>, <constant>INSTANCE_SECONDARIES</constant>, <constant>FORCE</constant></entry>
244 <entry>OP_INSTANCE_SHUTDOWN</entry>
245 <entry><filename class="directory">instance-shutdown</filename></entry>
246 <entry>Stops an instance</entry>
247 <entry><computeroutput>gnt-instance shutdown</computeroutput></entry>
248 <entry><constant>INSTANCE_NAME</constant>, <constant>INSTANCE_PRIMARY</constant>, <constant>INSTANCE_SECONDARIES</constant></entry>
251 <entry>OP_INSTANCE_MODIFY</entry>
252 <entry><filename class="directory">instance-modify</filename></entry>
253 <entry>Modifies the instance parameters.</entry>
254 <entry><computeroutput>gnt-instance modify</computeroutput></entry>
255 <entry><constant>INSTANCE_NAME</constant>, <constant>MEM_SIZE</constant>, <constant>VCPUS</constant>, <constant>INSTANCE_IP</constant></entry>
258 <entry>OP_INSTANCE_FAILOVER</entry>
259 <entry><filename class="directory">instance-failover</filename></entry>
260 <entry>Failover an instance</entry>
261 <entry><computeroutput>gnt-instance start</computeroutput></entry>
262 <entry><constant>INSTANCE_NAME</constant>, <constant>INSTANCE_PRIMARY</constant>, <constant>INSTANCE_SECONDARIES</constant>, <constant>IGNORE_CONSISTENCY</constant></entry>
265 <entry>OP_INSTANCE_REMOVE</entry>
266 <entry><filename class="directory">instance-remove</filename></entry>
267 <entry>Remove an instance</entry>
268 <entry><computeroutput>gnt-instance remove</computeroutput></entry>
269 <entry><constant>INSTANCE_NAME</constant>, <constant>INSTANCE_PRIMARY</constant>, <constant>INSTANCE_SECONDARIES</constant></entry>
270 <entry spanname="bothhooks">master node</entry>
273 <entry>OP_INSTANCE_ADD_MDDRBD</entry>
274 <entry><filename class="directory">mirror-add</filename></entry>
275 <entry>Adds a mirror component</entry>
276 <entry><computeroutput>gnt-instance add-mirror</computeroutput></entry>
277 <entry><constant>INSTANCE_NAME</constant>, <constant>NEW_SECONDARY</constant>, <constant>DISK_NAME</constant></entry>
280 <entry>OP_INSTANCE_REMOVE_MDDRBD</entry>
281 <entry><filename class="directory">mirror-remove</filename></entry>
282 <entry>Removes a mirror component</entry>
283 <entry><computeroutput>gnt-instance remove-mirror</computeroutput></entry>
284 <entry><constant>INSTANCE_NAME</constant>, <constant>OLD_SECONDARY</constant>, <constant>DISK_NAME</constant>, <constant>DISK_ID</constant></entry>
287 <entry>OP_INSTANCE_REPLACE_DISKS</entry>
288 <entry><filename class="directory">mirror-replace</filename></entry>
289 <entry>Replace all mirror components</entry>
290 <entry><computeroutput>gnt-instance replace-disks</computeroutput></entry>
291 <entry><constant>INSTANCE_NAME</constant>, <constant>OLD_SECONDARY</constant>, <constant>NEW_SECONDARY</constant></entry>
295 <entry>OP_CLUSTER_VERIFY</entry>
296 <entry><filename class="directory">cluster-verify</filename></entry>
297 <entry>Verifies the cluster status</entry>
298 <entry><computeroutput>gnt-cluster verify</computeroutput></entry>
299 <entry><constant>CLUSTER</constant>, <constant>MASTER</constant></entry>
301 <entry>all nodes</entry>
309 <title>Environment variables</title>
311 <para>Note that all variables listed here are actually prefixed
312 with <constant>GANETI_</constant> in order to provide a
313 different namespace.</para>
316 <title>Common variables</title>
318 <para>This is the list of environment variables supported by
319 all operations:</para>
323 <term>HOOKS_VERSION</term>
325 <para>Documents the hooks interface version. In case this
326 doesnt match what the script expects, it should not
327 run. The documents conforms to the version
328 <literal>1</literal>.</para>
332 <term>HOOKS_PHASE</term>
334 <para>one of <constant>PRE</constant> or
335 <constant>POST</constant> denoting which phase are we
342 <para>the cluster name</para>
348 <para>the master node</para>
354 <para>one of the <constant>OP_*</constant> values from
355 the table of operations</para>
359 <term>OBJECT_TYPE</term>
361 <para>one of <simplelist type="inline">
362 <member><constant>INSTANCE</constant></member>
363 <member><constant>NODE</constant></member>
364 <member><constant>CLUSTER</constant></member>
365 </simplelist>, showing the target of the operation.
369 <!-- commented out since it causes problems in our rpc
370 multi-node optimised calls
372 <term>HOST_NAME</term>
374 <para>The name of the node the hook is run on as known by
379 <term>HOST_TYPE</term>
381 <para>one of <simplelist type="inline">
382 <member><constant>MASTER</constant></member>
383 <member><constant>NODE</constant></member>
384 </simplelist>, showing the role of this node in the cluster.
393 <title>Specialised variables</title>
395 <para>This is the list of variables which are specific to one
396 or more operations.</para>
399 <term>INSTANCE_NAME</term>
401 <para>The name of the instance which is the target of
402 the operation.</para>
406 <term>INSTANCE_DISK_TYPE</term>
408 <para>The disk type for the instance.</para>
412 <term>INSTANCE_DISK_SIZE</term>
414 <para>The (OS) disk size for the instance.</para>
418 <term>INSTANCE_OS</term>
420 <para>The name of the instance OS.</para>
424 <term>INSTANCE_PRIMARY</term>
426 <para>The name of the node which is the primary for the
431 <term>INSTANCE_SECONDARIES</term>
433 <para>Space-separated list of secondary nodes for the
438 <term>NODE_NAME</term>
440 <para>The target node of this operation (not the node on
441 which the hook runs).</para>
445 <term>NODE_PIP</term>
447 <para>The primary IP of the target node (the one over
448 which inter-node communication is done).</para>
452 <term>NODE_SIP</term>
454 <para>The secondary IP of the target node (the one over
455 which drbd replication is done). This can be equal to
456 the primary ip, in case the cluster is not
461 <term>OLD_MASTER</term>
462 <term>NEW_MASTER</term>
464 <para>The old, respectively the new master for the
465 master failover operation.</para>
471 <para>This is provided by some operations when the user
472 gave this flag.</para>
476 <term>IGNORE_CONSISTENCY</term>
478 <para>The user has specified this flag. It is used when
479 failing over instances in case the primary node is
484 <term>MEM_SIZE, DISK_SIZE, SWAP_SIZE, VCPUS</term>
486 <para>The memory, disk, swap size and the number of
487 processor selected for the instance (in
488 <command>gnt-instance add</command> or
489 <command>gnt-instance modify</command>).</para>
493 <term>INSTANCE_IP</term>
495 <para>If defined, the instance IP in the
496 <command>gnt-instance add</command> and
497 <command>gnt-instance set</command> commands. If not
498 defined, it means that no IP has been defined.</para>
502 <term>DISK_TEMPLATE</term>
504 <para>The disk template type when creating the instance.</para>
508 <term>INSTANCE_ADD_MODE</term>
510 <para>The mode of the create: either
511 <constant>create</constant> for create from scratch or
512 <constant>import</constant> for restoring from an
513 exported image.</para>
517 <term>SRC_NODE, SRC_PATH, SRC_IMAGE</term>
519 <para>In case the instance has been added by import,
520 these variables are defined and point to the source
521 node, source path (the directory containing the image
522 and the config file) and the source disk image
527 <term>DISK_NAME</term>
529 <para>The disk name (either <filename>sda</filename> or
530 <filename>sdb</filename>) in mirror operations
531 (add/remove mirror).</para>
537 <para>The disk id for mirror remove operations. You can
538 look this up using <command>gnt-instance
539 info</command>.</para>
543 <term>NEW_SECONDARY</term>
545 <para>The name of the node on which the new mirror
546 componet is being added. This can be the name of the
547 current secondary, if the new mirror is on the same
552 <term>OLD_SECONDARY</term>
554 <para>The name of the old secondary. This is used in
555 both <command>replace-disks</command> and
556 <command>remove-mirror</command>. Note that this can be
557 equal to the new secondary (only
558 <command>replace-disks</command> has both variables) if
559 the secondary node hasn't actually changed).</para>
563 <term>EXPORT_NODE</term>
565 <para>The node on which the exported image of the
566 instance was done.</para>
570 <term>EXPORT_DO_SHUTDOWN</term>
572 <para>This variable tells if the instance has been
573 shutdown or not while doing the export. In the "was
574 shutdown" case, it's likely that the filesystem is
575 consistent, whereas in the "did not shutdown" case, the
576 filesystem would need a check (journal replay or full
577 fsck) in order to guarantee consistency.</para>