root / doc / hooks.rst @ d87e1814
History | View | Annotate | Download (14 kB)
1 |
Ganeti customisation using hooks |
---|---|
2 |
================================ |
3 |
|
4 |
Documents ganeti version 2.0 |
5 |
|
6 |
.. contents:: |
7 |
|
8 |
Introduction |
9 |
------------ |
10 |
|
11 |
|
12 |
In order to allow customisation of operations, ganeti runs scripts |
13 |
under ``/etc/ganeti/hooks`` based on certain rules. |
14 |
|
15 |
|
16 |
This is similar to the ``/etc/network/`` structure present in Debian |
17 |
for network interface handling. |
18 |
|
19 |
Organisation |
20 |
------------ |
21 |
|
22 |
For every operation, two sets of scripts are run: |
23 |
|
24 |
- pre phase (for authorization/checking) |
25 |
- post phase (for logging) |
26 |
|
27 |
Also, for each operation, the scripts are run on one or more nodes, |
28 |
depending on the operation type. |
29 |
|
30 |
Note that, even though we call them scripts, we are actually talking |
31 |
about any executable. |
32 |
|
33 |
*pre* scripts |
34 |
~~~~~~~~~~~~~ |
35 |
|
36 |
The *pre* scripts have a definite target: to check that the operation |
37 |
is allowed given the site-specific constraints. You could have, for |
38 |
example, a rule that says every new instance is required to exists in |
39 |
a database; to implement this, you could write a script that checks |
40 |
the new instance parameters against your database. |
41 |
|
42 |
The objective of these scripts should be their return code (zero or |
43 |
non-zero for success and failure). However, if they modify the |
44 |
environment in any way, they should be idempotent, as failed |
45 |
executions could be restarted and thus the script(s) run again with |
46 |
exactly the same parameters. |
47 |
|
48 |
Note that if a node is unreachable at the time a hooks is run, this |
49 |
will not be interpreted as a deny for the execution. In other words, |
50 |
only an actual error returned from a script will cause abort, and not |
51 |
an unreachable node. |
52 |
|
53 |
Therefore, if you want to guarantee that a hook script is run and |
54 |
denies an action, it's best to put it on the master node. |
55 |
|
56 |
*post* scripts |
57 |
~~~~~~~~~~~~~~ |
58 |
|
59 |
These scripts should do whatever you need as a reaction to the |
60 |
completion of an operation. Their return code is not checked (but |
61 |
logged), and they should not depend on the fact that the *pre* scripts |
62 |
have been run. |
63 |
|
64 |
Naming |
65 |
~~~~~~ |
66 |
|
67 |
The allowed names for the scripts consist of (similar to *run-parts* ) |
68 |
upper and lower case, digits, underscores and hyphens. In other words, |
69 |
the regexp ``^[a-zA-Z0-9_-]+$``. Also, non-executable scripts will be |
70 |
ignored. |
71 |
|
72 |
|
73 |
Order of execution |
74 |
~~~~~~~~~~~~~~~~~~ |
75 |
|
76 |
On a single node, the scripts in a directory are run in lexicographic |
77 |
order (more exactly, the python string comparison order). It is |
78 |
advisable to implement the usual *NN-name* convention where *NN* is a |
79 |
two digit number. |
80 |
|
81 |
For an operation whose hooks are run on multiple nodes, there is no |
82 |
specific ordering of nodes with regard to hooks execution; you should |
83 |
assume that the scripts are run in parallel on the target nodes |
84 |
(keeping on each node the above specified ordering). If you need any |
85 |
kind of inter-node synchronisation, you have to implement it yourself |
86 |
in the scripts. |
87 |
|
88 |
Execution environment |
89 |
~~~~~~~~~~~~~~~~~~~~~ |
90 |
|
91 |
The scripts will be run as follows: |
92 |
|
93 |
- no command line arguments |
94 |
|
95 |
- no controlling *tty* |
96 |
|
97 |
- stdin is actually */dev/null* |
98 |
|
99 |
- stdout and stderr are directed to files |
100 |
|
101 |
- PATH is reset to ``/sbin:/bin:/usr/sbin:/usr/bin`` |
102 |
|
103 |
- the environment is cleared, and only ganeti-specific variables will |
104 |
be left |
105 |
|
106 |
|
107 |
All information about the cluster is passed using environment |
108 |
variables. Different operations will have sligthly different |
109 |
environments, but most of the variables are common. |
110 |
|
111 |
Operation list |
112 |
-------------- |
113 |
|
114 |
Node operations |
115 |
~~~~~~~~~~~~~~~ |
116 |
|
117 |
OP_ADD_NODE |
118 |
+++++++++++ |
119 |
|
120 |
Adds a node to the cluster. |
121 |
|
122 |
:directory: node-add |
123 |
:env. vars: NODE_NAME, NODE_PIP, NODE_SIP |
124 |
:pre-execution: all existing nodes |
125 |
:post-execution: all nodes plus the new node |
126 |
|
127 |
|
128 |
OP_REMOVE_NODE |
129 |
++++++++++++++ |
130 |
|
131 |
Removes a node from the cluster. On the removed node the hooks are called |
132 |
during the execution of the operation and not after its completion. |
133 |
|
134 |
:directory: node-remove |
135 |
:env. vars: NODE_NAME |
136 |
:pre-execution: all existing nodes except the removed node |
137 |
:post-execution: all existing nodes |
138 |
|
139 |
OP_NODE_SET_PARAMS |
140 |
++++++++++++++++++ |
141 |
|
142 |
Changes a node's parameters. |
143 |
|
144 |
:directory: node-modify |
145 |
:env. vars: MASTER_CANDIDATE, OFFLINE, DRAINED |
146 |
:pre-execution: master node, the target node |
147 |
:post-execution: master node, the target node |
148 |
|
149 |
OP_NODE_EVACUATE |
150 |
++++++++++++++++ |
151 |
|
152 |
Relocate secondary instances from a node. |
153 |
|
154 |
:directory: node-evacuate |
155 |
:env. vars: NEW_SECONDARY, NODE_NAME |
156 |
:pre-execution: master node, target node |
157 |
:post-execution: master node, target node |
158 |
|
159 |
|
160 |
Instance operations |
161 |
~~~~~~~~~~~~~~~~~~~ |
162 |
|
163 |
All instance operations take at least the following variables: |
164 |
INSTANCE_NAME, INSTANCE_PRIMARY, INSTANCE_SECONDARIES, |
165 |
INSTANCE_OS_TYPE, INSTANCE_DISK_TEMPLATE, INSTANCE_MEMORY, |
166 |
INSTANCE_DISK_SIZES, INSTANCE_VCPUS, INSTANCE_NIC_COUNT, |
167 |
INSTANCE_NICn_IP, INSTANCE_NICn_BRIDGE, INSTANCE_NICn_MAC, |
168 |
INSTANCE_DISK_COUNT, INSTANCE_DISKn_SIZE, INSTANCE_DISKn_MODE. |
169 |
|
170 |
The INSTANCE_NICn_* and INSTANCE_DISKn_* variables represent the |
171 |
properties of the *n* -th NIC and disk, and are zero-indexed. |
172 |
|
173 |
|
174 |
OP_INSTANCE_ADD |
175 |
+++++++++++++++ |
176 |
|
177 |
Creates a new instance. |
178 |
|
179 |
:directory: instance-add |
180 |
:env. vars: ADD_MODE, SRC_NODE, SRC_PATH, SRC_IMAGES |
181 |
:pre-execution: master node, primary and secondary nodes |
182 |
:post-execution: master node, primary and secondary nodes |
183 |
|
184 |
OP_INSTANCE_REINSTALL |
185 |
+++++++++++++++++++++ |
186 |
|
187 |
Reinstalls an instance. |
188 |
|
189 |
:directory: instance-reinstall |
190 |
:env. vars: only the standard instance vars |
191 |
:pre-execution: master node, primary and secondary nodes |
192 |
:post-execution: master node, primary and secondary nodes |
193 |
|
194 |
OP_BACKUP_EXPORT |
195 |
++++++++++++++++ |
196 |
|
197 |
Exports the instance. |
198 |
|
199 |
|
200 |
:directory: instance-export |
201 |
:env. vars: EXPORT_NODE, EXPORT_DO_SHUTDOWN |
202 |
:pre-execution: master node, primary and secondary nodes |
203 |
:post-execution: master node, primary and secondary nodes |
204 |
|
205 |
OP_INSTANCE_START |
206 |
+++++++++++++++++ |
207 |
|
208 |
Starts an instance. |
209 |
|
210 |
:directory: instance-start |
211 |
:env. vars: INSTANCE_NAME, INSTANCE_PRIMARY, INSTANCE_SECONDARIES, FORCE |
212 |
:pre-execution: master node, primary and secondary nodes |
213 |
:post-execution: master node, primary and secondary nodes |
214 |
|
215 |
OP_INSTANCE_SHUTDOWN |
216 |
++++++++++++++++++++ |
217 |
|
218 |
Stops an instance. |
219 |
|
220 |
:directory: instance-shutdown |
221 |
:env. vars: INSTANCE_NAME, INSTANCE_PRIMARY, INSTANCE_SECONDARIES |
222 |
:pre-execution: master node, primary and secondary nodes |
223 |
:post-execution: master node, primary and secondary nodes |
224 |
|
225 |
OP_INSTANCE_REBOOT |
226 |
++++++++++++++++++ |
227 |
|
228 |
Reboots an instance. |
229 |
|
230 |
:directory: instance-reboot |
231 |
:env. vars: IGNORE_SECONDARIES, REBOOT_TYPE |
232 |
:pre-execution: master node, primary and secondary nodes |
233 |
:post-execution: master node, primary and secondary nodes |
234 |
|
235 |
OP_INSTANCE_MODIFY |
236 |
++++++++++++++++++ |
237 |
|
238 |
Modifies the instance parameters. |
239 |
|
240 |
:directory: instance-modify |
241 |
:env. vars: INSTANCE_NAME, MEM_SIZE, VCPUS, INSTANCE_IP |
242 |
:pre-execution: master node, primary and secondary nodes |
243 |
:post-execution: master node, primary and secondary nodes |
244 |
|
245 |
OP_INSTANCE_FAILOVER |
246 |
++++++++++++++++++++ |
247 |
|
248 |
Failovers an instance. |
249 |
|
250 |
:directory: instance-failover |
251 |
:env. vars: IGNORE_CONSISTENCY |
252 |
:pre-execution: master node, secondary node |
253 |
:post-execution: master node, secondary node |
254 |
|
255 |
OP_INSTANCE_MIGRATE |
256 |
++++++++++++++++++++ |
257 |
|
258 |
Migrates an instance. |
259 |
|
260 |
:directory: instance-failover |
261 |
:env. vars: INSTANCE_MIGRATE_LIVE, INSTANCE_MIGRATE_CLEANUP |
262 |
:pre-execution: master node, secondary node |
263 |
:post-execution: master node, secondary node |
264 |
|
265 |
|
266 |
OP_INSTANCE_REMOVE |
267 |
++++++++++++++++++ |
268 |
|
269 |
Remove an instance. |
270 |
|
271 |
:directory: instance-remove |
272 |
:env. vars: INSTANCE_NAME, INSTANCE_PRIMARY, INSTANCE_SECONDARIES |
273 |
:pre-execution: master node |
274 |
:post-execution: master node |
275 |
|
276 |
OP_INSTANCE_REPLACE_DISKS |
277 |
+++++++++++++++++++++++++ |
278 |
|
279 |
Replace an instance's disks. |
280 |
|
281 |
:directory: mirror-replace |
282 |
:env. vars: MODE, NEW_SECONDARY, OLD_SECONDARY |
283 |
:pre-execution: master node, primary and secondary nodes |
284 |
:post-execution: master node, primary and secondary nodes |
285 |
|
286 |
OP_INSTANCE_GROW_DISK |
287 |
+++++++++++++++++++++ |
288 |
|
289 |
Grows the disk of an instance. |
290 |
|
291 |
:directory: disk-grow |
292 |
:env. vars: DISK, AMOUNT |
293 |
:pre-execution: master node, primary node |
294 |
:post-execution: master node, primary node |
295 |
|
296 |
OP_INSTANCE_RENAME |
297 |
++++++++++++++++++ |
298 |
|
299 |
Renames an instance. |
300 |
|
301 |
:directory: instance-rename |
302 |
:env. vars: INSTANCE_NEW_NAME |
303 |
:pre-execution: master node, primary and secondary nodes |
304 |
:post-execution: master node, primary and secondary nodes |
305 |
|
306 |
Cluster operations |
307 |
~~~~~~~~~~~~~~~~~~ |
308 |
|
309 |
OP_POST_INIT_CLUSTER |
310 |
++++++++++++++++++++ |
311 |
|
312 |
This hook is called via a special "empty" LU right after cluster initialization. |
313 |
|
314 |
:directory: cluster-init |
315 |
:env. vars: none |
316 |
:pre-execution: none |
317 |
:post-execution: master node |
318 |
|
319 |
OP_DESTROY_CLUSTER |
320 |
++++++++++++++++++ |
321 |
|
322 |
The post phase of this hook is called during the execution of destroy operation |
323 |
and not after its completion. |
324 |
|
325 |
:directory: cluster-destroy |
326 |
:env. vars: none |
327 |
:pre-execution: none |
328 |
:post-execution: master node |
329 |
|
330 |
OP_CLUSTER_VERIFY |
331 |
+++++++++++++++++ |
332 |
|
333 |
Verifies the cluster status. This is a special LU with regard to |
334 |
hooks, as the result of the opcode will be combined with the result of |
335 |
post-execution hooks, in order to allow administrators to enhance the |
336 |
cluster verification procedure. |
337 |
|
338 |
:directory: cluster-verify |
339 |
:env. vars: CLUSTER, MASTER, CLUSTER_TAGS, NODE_TAGS_<name> |
340 |
:pre-execution: none |
341 |
:post-execution: all nodes |
342 |
|
343 |
OP_CLUSTER_RENAME |
344 |
+++++++++++++++++ |
345 |
|
346 |
Renames the cluster. |
347 |
|
348 |
:directory: cluster-rename |
349 |
:env. vars: NEW_NAME |
350 |
:pre-execution: master-node |
351 |
:post-execution: master-node |
352 |
|
353 |
OP_CLUSTER_SET_PARAMS |
354 |
+++++++++++++++++++++ |
355 |
|
356 |
Modifies the cluster parameters. |
357 |
|
358 |
:directory: cluster-modify |
359 |
:env. vars: NEW_VG_NAME |
360 |
:pre-execution: master node |
361 |
:post-execution: master node |
362 |
|
363 |
|
364 |
Obsolete operations |
365 |
~~~~~~~~~~~~~~~~~~~ |
366 |
|
367 |
The following operations are no longer present or don't execute hooks |
368 |
anymore in Ganeti 2.0: |
369 |
|
370 |
- OP_INIT_CLUSTER |
371 |
- OP_MASTER_FAILOVER |
372 |
- OP_INSTANCE_ADD_MDDRBD |
373 |
- OP_INSTANCE_REMOVE_MDDRBD |
374 |
|
375 |
|
376 |
Environment variables |
377 |
--------------------- |
378 |
|
379 |
Note that all variables listed here are actually prefixed with |
380 |
*GANETI_* in order to provide a clear namespace. |
381 |
|
382 |
Common variables |
383 |
~~~~~~~~~~~~~~~~ |
384 |
|
385 |
This is the list of environment variables supported by all operations: |
386 |
|
387 |
HOOKS_VERSION |
388 |
Documents the hooks interface version. In case this doesnt match |
389 |
what the script expects, it should not run. The documents conforms |
390 |
to the version 2. |
391 |
|
392 |
HOOKS_PHASE |
393 |
One of *PRE* or *POST* denoting which phase are we in. |
394 |
|
395 |
CLUSTER |
396 |
The cluster name. |
397 |
|
398 |
MASTER |
399 |
The master node. |
400 |
|
401 |
OP_CODE |
402 |
One of the *OP_* values from the list of operations. |
403 |
|
404 |
OBJECT_TYPE |
405 |
One of ``INSTANCE``, ``NODE``, ``CLUSTER``. |
406 |
|
407 |
DATA_DIR |
408 |
The path to the Ganeti configuration directory (to read, for |
409 |
example, the *ssconf* files). |
410 |
|
411 |
|
412 |
Specialised variables |
413 |
~~~~~~~~~~~~~~~~~~~~~ |
414 |
|
415 |
This is the list of variables which are specific to one or more |
416 |
operations. |
417 |
|
418 |
INSTANCE_NAME |
419 |
The name of the instance which is the target of the operation. |
420 |
|
421 |
INSTANCE_DISK_TEMPLATE |
422 |
The disk type for the instance. |
423 |
|
424 |
INSTANCE_DISK_COUNT |
425 |
The number of disks for the instance. |
426 |
|
427 |
INSTANCE_DISKn_SIZE |
428 |
The size of disk *n* for the instance. |
429 |
|
430 |
INSTANCE_DISKn_MODE |
431 |
Either *rw* for a read-write disk or *ro* for a read-only one. |
432 |
|
433 |
INSTANCE_NIC_COUNT |
434 |
The number of NICs for the instance. |
435 |
|
436 |
INSTANCE_NICn_BRIDGE |
437 |
The bridge to which the *n* -th NIC of the instance is attached. |
438 |
|
439 |
INSTANCE_NICn_IP |
440 |
The IP (if any) of the *n* -th NIC of the instance. |
441 |
|
442 |
INSTANCE_NICn_MAC |
443 |
The MAC address of the *n* -th NIC of the instance. |
444 |
|
445 |
INSTANCE_OS_TYPE |
446 |
The name of the instance OS. |
447 |
|
448 |
INSTANCE_PRIMARY |
449 |
The name of the node which is the primary for the instance. |
450 |
|
451 |
INSTANCE_SECONDARIES |
452 |
Space-separated list of secondary nodes for the instance. |
453 |
|
454 |
INSTANCE_MEMORY |
455 |
The memory size (in MiBs) of the instance. |
456 |
|
457 |
INSTANCE_VCPUS |
458 |
The number of virtual CPUs for the instance. |
459 |
|
460 |
INSTANCE_STATUS |
461 |
The run status of the instance. |
462 |
|
463 |
NODE_NAME |
464 |
The target node of this operation (not the node on which the hook |
465 |
runs). |
466 |
|
467 |
NODE_PIP |
468 |
The primary IP of the target node (the one over which inter-node |
469 |
communication is done). |
470 |
|
471 |
NODE_SIP |
472 |
The secondary IP of the target node (the one over which drbd |
473 |
replication is done). This can be equal to the primary ip, in case |
474 |
the cluster is not dual-homed. |
475 |
|
476 |
FORCE |
477 |
This is provided by some operations when the user gave this flag. |
478 |
|
479 |
IGNORE_CONSISTENCY |
480 |
The user has specified this flag. It is used when failing over |
481 |
instances in case the primary node is down. |
482 |
|
483 |
ADD_MODE |
484 |
The mode of the instance create: either *create* for create from |
485 |
scratch or *import* for restoring from an exported image. |
486 |
|
487 |
SRC_NODE, SRC_PATH, SRC_IMAGE |
488 |
In case the instance has been added by import, these variables are |
489 |
defined and point to the source node, source path (the directory |
490 |
containing the image and the config file) and the source disk image |
491 |
file. |
492 |
|
493 |
NEW_SECONDARY |
494 |
The name of the node on which the new mirror component is being |
495 |
added. This can be the name of the current secondary, if the new |
496 |
mirror is on the same secondary. |
497 |
|
498 |
OLD_SECONDARY |
499 |
The name of the old secondary in the replace-disks command Note that |
500 |
this can be equal to the new secondary if the secondary node hasn't |
501 |
actually changed. |
502 |
|
503 |
EXPORT_NODE |
504 |
The node on which the exported image of the instance was done. |
505 |
|
506 |
EXPORT_DO_SHUTDOWN |
507 |
This variable tells if the instance has been shutdown or not while |
508 |
doing the export. In the "was shutdown" case, it's likely that the |
509 |
filesystem is consistent, whereas in the "did not shutdown" case, |
510 |
the filesystem would need a check (journal replay or full fsck) in |
511 |
order to guarantee consistency. |
512 |
|
513 |
CLUSTER_TAGS |
514 |
The list of cluster tags, space separated. |
515 |
|
516 |
NODE_TAGS_<name> |
517 |
The list of tags for node *<name>*, space separated. |
518 |
|
519 |
Examples |
520 |
-------- |
521 |
|
522 |
The startup of an instance will pass this environment to the hook |
523 |
script:: |
524 |
|
525 |
GANETI_CLUSTER=cluster1.example.com |
526 |
GANETI_DATA_DIR=/var/lib/ganeti |
527 |
GANETI_FORCE=False |
528 |
GANETI_HOOKS_PATH=instance-start |
529 |
GANETI_HOOKS_PHASE=post |
530 |
GANETI_HOOKS_VERSION=2 |
531 |
GANETI_INSTANCE_DISK0_MODE=rw |
532 |
GANETI_INSTANCE_DISK0_SIZE=128 |
533 |
GANETI_INSTANCE_DISK_COUNT=1 |
534 |
GANETI_INSTANCE_DISK_TEMPLATE=drbd |
535 |
GANETI_INSTANCE_MEMORY=128 |
536 |
GANETI_INSTANCE_NAME=instance2.example.com |
537 |
GANETI_INSTANCE_NIC0_BRIDGE=xen-br0 |
538 |
GANETI_INSTANCE_NIC0_IP= |
539 |
GANETI_INSTANCE_NIC0_MAC=aa:00:00:a5:91:58 |
540 |
GANETI_INSTANCE_NIC_COUNT=1 |
541 |
GANETI_INSTANCE_OS_TYPE=debootstrap |
542 |
GANETI_INSTANCE_PRIMARY=node3.example.com |
543 |
GANETI_INSTANCE_SECONDARIES=node5.example.com |
544 |
GANETI_INSTANCE_STATUS=down |
545 |
GANETI_INSTANCE_VCPUS=1 |
546 |
GANETI_MASTER=node1.example.com |
547 |
GANETI_OBJECT_TYPE=INSTANCE |
548 |
GANETI_OP_CODE=OP_INSTANCE_STARTUP |
549 |
GANETI_OP_TARGET=instance2.example.com |