1 Ganeti automatic instance allocation
2 ====================================
4 Documents Ganeti version 2.4
11 Currently in Ganeti the admin has to specify the exact locations for
12 an instance's node(s). This prevents a completely automatic node
13 evacuation, and is in general a nuisance.
15 The *iallocator* framework will enable automatic placement via
16 external scripts, which allows customization of the cluster layout per
17 the site's requirements.
22 There are two parts of the ganeti operation that are impacted by the
23 auto-allocation: how the cluster knows what the allocator algorithms
24 are and how the admin uses these in creating instances.
26 An allocation algorithm is just the filename of a program installed in
27 a defined list of directories.
32 At configure time, the list of the directories can be selected via the
33 ``--with-iallocator-search-path=LIST`` option, where *LIST* is a
34 comma-separated list of directories. If not given, this defaults to
35 ``$libdir/ganeti/iallocators``, i.e. for an installation under
36 ``/usr``, this will be ``/usr/lib/ganeti/iallocators``.
38 Ganeti will then search for allocator script in the configured list,
39 using the first one whose filename matches the one given by the user.
41 Command line interface changes
42 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
44 The node selection options in instanece add and instance replace disks
45 can be replace by the new ``--iallocator=NAME`` option (shortened to
46 ``-I``), which will cause the auto-assignement of nodes with the
47 passed iallocator. The selected node(s) will be show as part of the
53 The protocol for communication between Ganeti and an allocator script
54 will be the following:
56 #. ganeti launches the program with a single argument, a filename that
57 contains a JSON-encoded structure (the input message)
59 #. if the script finishes with exit code different from zero, it is
60 considered a general failure and the full output will be reported to
61 the users; this can be the case when the allocator can't parse the
64 #. if the allocator finishes with exit code zero, it is expected to
65 output (on its stdout) a JSON-encoded structure (the response)
70 The input message will be the JSON encoding of a dictionary containing
71 all the required information to perform the operation. We explain the
72 contents of this dictionary in two parts: common information that every
73 type of operation requires, and operation-specific information.
78 All input dictionaries to the IAllocator must carry the following keys:
81 the version of the protocol; this document
88 the list of cluster tags
91 the list of enabled hypervisors
94 a dictionary containing the details of the request; the keys vary
95 depending on the type of operation that's being requested, as
96 explained in `Operation-specific input`_ below.
99 a dictionary with the data for the cluster's node groups; it is keyed
100 on the group UUID, and the values are a dictionary with the following
106 the allocation policy of the node group (consult the semantics of
107 this attribute in the :manpage:`gnt-group(8)` manpage)
110 a dictionary with the data for the current existing instance on the
111 cluster, indexed by instance name; the contents are similar to the
112 instance definitions for the allocate mode, with the addition of:
115 if this instance is set to run (but not the actual status of the
119 list of nodes on which this instance is placed; the primary node
120 of the instance is always the first one
123 dictionary with the data for the nodes in the cluster, indexed by
124 the node name; the dict contains [*]_ :
127 the total disk size of this node (mebibytes)
130 the free disk space on the node
133 the total memory size
136 free memory on the node; note that currently this does not take
137 into account the instances which are down on the node
140 the physical number of CPUs present on the machine; depending on
141 the hypervisor, this might or might not be equal to how many CPUs
142 the node operating system sees;
145 the primary IP address of the node
148 the secondary IP address of the node (the one used for the DRBD
149 replication); note that this can be the same as the primary one
152 list with the tags of the node
155 a boolean flag denoting whether this node is a master candidate
158 a boolean flag denoting whether this node is being drained
161 a boolean flag denoting whether this node is offline
164 total memory required by primary instances
167 total memory required by running primary instances
170 the node group that this node belongs to
172 No allocations should be made on nodes having either the ``drained``
173 or ``offline`` flags set. More details about these of node status
174 flags is available in the manpage :manpage:`ganeti(7)`.
176 .. [*] Note that no run-time data is present for offline, drained or
177 non-vm_capable nodes; this means the tags total_memory,
178 reserved_memory, free_memory, total_disk, free_disk, total_cpus,
179 i_pri_memory and i_pri_up memory will be absent
181 Operation-specific input
182 ++++++++++++++++++++++++
184 All input dictionaries to the IAllocator carry, in the ``request``
185 dictionary, detailed information about the operation that's being
186 requested. The required keys vary depending on the type of operation, as
189 In all cases, it includes:
192 the request type; this can be either ``allocate``, ``relocate``,
193 ``change-group``, ``node-evacuate`` or ``multi-evacuate``. The
194 ``allocate`` request is used when a new instance needs to be placed
195 on the cluster. The ``relocate`` request is used when an existing
196 instance needs to be moved within its node group.
198 The ``multi-evacuate`` protocol used to request that the script
199 computes the optimal relocate solution for all secondary instances
200 of the given nodes. It is now deprecated and should no longer be
203 The ``change-group`` request is used to relocate multiple instances
204 across multiple node groups. ``node-evacuate`` evacuates instances
205 off their node(s). These are described in a separate :ref:`design
206 document <multi-reloc-detailed-design>`.
208 For both allocate and relocate mode, the following extra keys are needed
209 in the ``request`` dictionary:
212 the name of the instance; if the request is a realocation, then this
213 name will be found in the list of instances (see below), otherwise
214 is the FQDN of the new instance; type *string*
217 how many nodes should the algorithm return; while this information
218 can be deduced from the instace's disk template, it's better if
219 this computation is left to Ganeti as then allocator scripts are
220 less sensitive to changes to the disk templates; type *integer*
223 the total disk space that will be used by this instance on the
224 (new) nodes; again, this information can be computed from the list
225 of instance disks and its template type, but Ganeti is better
226 suited to compute it; type *integer*
230 constants.DISK_ACCESS_SET == set([constants.DISK_RDONLY,
231 constants.DISK_RDWR])
233 Allocation needs, in addition:
236 list of dictionaries holding the disk definitions for this
237 instance (in the order they are exported to the hypervisor):
240 either :pyeval:`constants.DISK_RDONLY` or
241 :pyeval:`constants.DISK_RDWR` denoting if the disk is read-only or
245 the size of this disk in mebibytes
248 a list of dictionaries holding the network interfaces for this
249 instance, containing:
252 the IP address that Ganeti know for this instance, or null
255 the MAC address for this interface
258 the bridge to which this interface will be connected
261 the number of VCPUs for the instance
264 the disk template for the instance
267 the memory size for the instance
270 the OS type for the instance
273 the list of the instance's tags
276 the hypervisor of this instance
281 a list of nodes to move the instance away from (note that with
282 Ganeti 2.0, this list will always contain a single node, the
283 current secondary of the instance); type *list of strings*
285 As for ``node-evacuate``, it needs the following request arguments:
288 a list of instance names to evacuate; type *list of strings*
291 specify which instances to evacuate; one of ``primary-only``,
292 ``secondary-only``, ``all``, type *string*
294 ``change-group`` needs the following request arguments:
297 a list of instance names whose group to change; type
301 must either be the empty list, or contain a list of group UUIDs that
302 should be considered for relocating instances to; type
305 Finally, in the case of multi-evacuate, there's one single request
306 argument (in addition to ``type``):
309 the names of the nodes to be evacuated; type *list of strings*
314 The response message is much more simple than the input one. It is
315 also a dict having three keys:
318 a boolean value denoting if the allocation was successful or not
321 a string with information from the scripts; if the allocation fails,
322 this will be shown to the user
325 the output of the algorithm; even if the algorithm failed
326 (i.e. success is false), this must be returned as an empty list
328 for allocate/relocate, this is the list of node(s) for the instance;
329 note that the length of this list must equal the ``requested_nodes``
330 entry in the input message, otherwise Ganeti will consider the result
333 for the ``node-evacuate`` and ``change-group`` modes, this is a
334 dictionary containing, among other information, a list of lists of
335 serialized opcodes; see the :ref:`design document
336 <multi-reloc-result>` for a detailed description
338 for multi-evacuation mode, this is a list of lists; each element of
339 the list is a list of instance name and the new secondary node
341 .. note:: Current Ganeti version accepts either ``result`` or ``nodes``
342 as a backwards-compatibility measure (older versions only supported
348 Input messages to scripts
349 ~~~~~~~~~~~~~~~~~~~~~~~~~
351 Input message, new instance allocation (common elements are listed this
352 time, but not included in further examples below)::
356 "cluster_name": "cluster1.example.com",
358 "enabled_hypervisors": [
362 "f4e06e0d-528a-4963-a5ad-10f3e114232d": {
364 "alloc_policy": "preferred"
368 "instance1.example.com": {
384 "mac": "aa:00:00:00:60:bf",
389 "disk_template": "plain",
394 "os": "debootstrap+default"
396 "instance2.example.com": {
412 "mac": "aa:00:00:55:f8:38",
417 "disk_template": "drbd",
423 "os": "debootstrap+default"
427 "node1.example.com": {
428 "total_disk": 858276,
429 "primary_ip": "198.51.100.1",
430 "secondary_ip": "192.0.2.1",
432 "group": "f4e06e0d-528a-4963-a5ad-10f3e114232d",
437 "node2.example.com": {
438 "total_disk": 858240,
439 "primary_ip": "198.51.100.2",
440 "secondary_ip": "192.0.2.2",
442 "group": "f4e06e0d-528a-4963-a5ad-10f3e114232d",
447 "node3.example.com.com": {
448 "total_disk": 572184,
449 "primary_ip": "198.51.100.3",
450 "secondary_ip": "192.0.2.3",
452 "group": "f4e06e0d-528a-4963-a5ad-10f3e114232d",
460 "name": "instance3.example.com",
462 "disk_space_total": 3328,
476 "mac": "00:11:22:33:44:55",
481 "disk_template": "drbd",
483 "os": "debootstrap+default",
488 hypervisor: "xen-pvm"
492 Input message, reallocation::
499 "name": "instance2.example.com",
501 "disk_space_total": 832,
508 Input message, node evacuation::
514 "type": "multi-evacuate",
524 Successful response message::
528 "info": "Allocation successful",
535 Failed response message::
539 "info": "Can't find a suitable node for position 2 (already selected: node2.example.com)",
543 Successful node evacuation message::
547 "info": "Request successful",
561 Command line messages
562 ~~~~~~~~~~~~~~~~~~~~~
565 # gnt-instance add -t plain -m 2g --os-size 1g --swap-size 512m --iallocator hail -o debootstrap+default instance3
566 Selected nodes for the instance: node1.example.com
567 * creating instance disks...
570 # gnt-instance add -t plain -m 3400m --os-size 1g --swap-size 512m --iallocator hail -o debootstrap+default instance4
571 Failure: prerequisites not met for this operation:
572 Can't compute nodes using iallocator 'hail': Can't find a suitable node for position 1 (already selected: )
574 # gnt-instance add -t drbd -m 1400m --os-size 1g --swap-size 512m --iallocator hail -o debootstrap+default instance5
575 Failure: prerequisites not met for this operation:
576 Can't compute nodes using iallocator 'hail': Can't find a suitable node for position 2 (already selected: node1.example.com)
578 Reference implementation
579 ~~~~~~~~~~~~~~~~~~~~~~~~
581 Ganeti's default iallocator is "hail" which is available when "htools"
582 components have been enabled at build time (see :doc:`install-quick` for
585 .. vim: set textwidth=72 :