]>
Commit | Line | Data |
---|---|---|
80c0adcb | 1 | [[chapter_ha_manager]] |
22653ac8 | 2 | ifdef::manvolnum[] |
b2f242ab DM |
3 | ha-manager(1) |
4 | ============= | |
5f09af76 DM |
5 | :pve-toplevel: |
6 | ||
22653ac8 DM |
7 | NAME |
8 | ---- | |
9 | ||
734404b4 | 10 | ha-manager - Proxmox VE HA Manager |
22653ac8 | 11 | |
49a5e11c | 12 | SYNOPSIS |
22653ac8 DM |
13 | -------- |
14 | ||
15 | include::ha-manager.1-synopsis.adoc[] | |
16 | ||
17 | DESCRIPTION | |
18 | ----------- | |
19 | endif::manvolnum[] | |
22653ac8 DM |
20 | ifndef::manvolnum[] |
21 | High Availability | |
22 | ================= | |
5f09af76 | 23 | :pve-toplevel: |
194d2f29 | 24 | endif::manvolnum[] |
b5266e9f DM |
25 | |
26 | Our modern society depends heavily on information provided by | |
27 | computers over the network. Mobile devices amplified that dependency, | |
28 | because people can access the network any time from anywhere. If you | |
29 | provide such services, it is very important that they are available | |
30 | most of the time. | |
31 | ||
32 | We can mathematically define the availability as the ratio of (A) the | |
33 | total time a service is capable of being used during a given interval | |
34 | to (B) the length of the interval. It is normally expressed as a | |
35 | percentage of uptime in a given year. | |
36 | ||
37 | .Availability - Downtime per Year | |
38 | [width="60%",cols="<d,d",options="header"] | |
39 | |=========================================================== | |
40 | |Availability % |Downtime per year | |
41 | |99 |3.65 days | |
42 | |99.9 |8.76 hours | |
43 | |99.99 |52.56 minutes | |
44 | |99.999 |5.26 minutes | |
45 | |99.9999 |31.5 seconds | |
46 | |99.99999 |3.15 seconds | |
47 | |=========================================================== | |
48 | ||
04bde502 DM |
49 | There are several ways to increase availability. The most elegant |
50 | solution is to rewrite your software, so that you can run it on | |
51 | several host at the same time. The software itself need to have a way | |
2af6af05 | 52 | to detect errors and do failover. This is relatively easy if you just |
04bde502 DM |
53 | want to serve read-only web pages. But in general this is complex, and |
54 | sometimes impossible because you cannot modify the software | |
55 | yourself. The following solutions works without modifying the | |
56 | software: | |
57 | ||
8c1189b6 | 58 | * Use reliable ``server'' components |
fd9e8984 | 59 | + |
04bde502 | 60 | NOTE: Computer components with same functionality can have varying |
2af6af05 | 61 | reliability numbers, depending on the component quality. Most vendors |
8c1189b6 | 62 | sell components with higher reliability as ``server'' components - |
04bde502 | 63 | usually at higher price. |
b5266e9f DM |
64 | |
65 | * Eliminate single point of failure (redundant components) | |
8c1189b6 FG |
66 | ** use an uninterruptible power supply (UPS) |
67 | ** use redundant power supplies on the main boards | |
68 | ** use ECC-RAM | |
69 | ** use redundant network hardware | |
70 | ** use RAID for local storage | |
71 | ** use distributed, redundant storage for VM data | |
b5266e9f DM |
72 | |
73 | * Reduce downtime | |
8c1189b6 FG |
74 | ** rapidly accessible administrators (24/7) |
75 | ** availability of spare parts (other nodes in a {pve} cluster) | |
76 | ** automatic error detection (provided by `ha-manager`) | |
77 | ** automatic failover (provided by `ha-manager`) | |
b5266e9f | 78 | |
5771d9b0 | 79 | Virtualization environments like {pve} make it much easier to reach |
8c1189b6 | 80 | high availability because they remove the ``hardware'' dependency. They |
04bde502 DM |
81 | also support to setup and use redundant storage and network |
82 | devices. So if one host fail, you can simply start those services on | |
43da8322 DM |
83 | another host within your cluster. |
84 | ||
8c1189b6 | 85 | Even better, {pve} provides a software stack called `ha-manager`, |
43da8322 DM |
86 | which can do that automatically for you. It is able to automatically |
87 | detect errors and do automatic failover. | |
88 | ||
8c1189b6 | 89 | {pve} `ha-manager` works like an ``automated'' administrator. First, you |
43da8322 | 90 | configure what resources (VMs, containers, ...) it should |
8c1189b6 FG |
91 | manage. `ha-manager` then observes correct functionality, and handles |
92 | service failover to another node in case of errors. `ha-manager` can | |
43da8322 DM |
93 | also handle normal user requests which may start, stop, relocate and |
94 | migrate a service. | |
04bde502 DM |
95 | |
96 | But high availability comes at a price. High quality components are | |
97 | more expensive, and making them redundant duplicates the costs at | |
98 | least. Additional spare parts increase costs further. So you should | |
99 | carefully calculate the benefits, and compare with those additional | |
100 | costs. | |
101 | ||
102 | TIP: Increasing availability from 99% to 99.9% is relatively | |
103 | simply. But increasing availability from 99.9999% to 99.99999% is very | |
8c1189b6 | 104 | hard and costly. `ha-manager` has typical error detection and failover |
43da8322 DM |
105 | times of about 2 minutes, so you can get no more than 99.999% |
106 | availability. | |
b5266e9f | 107 | |
823fa863 | 108 | |
5bd515d4 DM |
109 | Requirements |
110 | ------------ | |
3810ae1e | 111 | |
823fa863 DM |
112 | You must meet the following requirements before you start with HA: |
113 | ||
5bd515d4 | 114 | * at least three cluster nodes (to get reliable quorum) |
43da8322 | 115 | |
5bd515d4 | 116 | * shared storage for VMs and containers |
43da8322 | 117 | |
5bd515d4 | 118 | * hardware redundancy (everywhere) |
3810ae1e | 119 | |
823fa863 DM |
120 | * use reliable “server” components |
121 | ||
5bd515d4 | 122 | * hardware watchdog - if not available we fall back to the |
8c1189b6 | 123 | linux kernel software watchdog (`softdog`) |
3810ae1e | 124 | |
5bd515d4 | 125 | * optional hardware fencing devices |
3810ae1e | 126 | |
3810ae1e | 127 | |
80c0adcb | 128 | [[ha_manager_resources]] |
5bd515d4 DM |
129 | Resources |
130 | --------- | |
131 | ||
8c1189b6 FG |
132 | We call the primary management unit handled by `ha-manager` a |
133 | resource. A resource (also called ``service'') is uniquely | |
5bd515d4 | 134 | identified by a service ID (SID), which consists of the resource type |
8c1189b6 FG |
135 | and an type specific ID, e.g.: `vm:100`. That example would be a |
136 | resource of type `vm` (virtual machine) with the ID 100. | |
5bd515d4 DM |
137 | |
138 | For now we have two important resources types - virtual machines and | |
139 | containers. One basic idea here is that we can bundle related software | |
140 | into such VM or container, so there is no need to compose one big | |
8c1189b6 | 141 | service from other services, like it was done with `rgmanager`. In |
4c34defd | 142 | general, a HA managed resource should not depend on other resources. |
3810ae1e | 143 | |
22653ac8 | 144 | |
2b52e195 | 145 | How It Works |
22653ac8 DM |
146 | ------------ |
147 | ||
c7470421 DM |
148 | This section provides a detailed description of the {PVE} HA manager |
149 | internals. It describes all involved daemons and how they work | |
150 | together. To provide HA, two daemons run on each node: | |
3810ae1e | 151 | |
8c1189b6 | 152 | `pve-ha-lrm`:: |
3810ae1e | 153 | |
1600c60a DM |
154 | The local resource manager (LRM), which controls the services running on |
155 | the local node. It reads the requested states for its services from | |
156 | the current manager status file and executes the respective commands. | |
3810ae1e | 157 | |
8c1189b6 | 158 | `pve-ha-crm`:: |
3810ae1e | 159 | |
1600c60a DM |
160 | The cluster resource manager (CRM), which makes the cluster wide |
161 | decisions. It sends commands to the LRM, processes the results, | |
162 | and moves resources to other nodes if something fails. The CRM also | |
163 | handles node fencing. | |
164 | ||
3810ae1e TL |
165 | |
166 | .Locks in the LRM & CRM | |
167 | [NOTE] | |
168 | Locks are provided by our distributed configuration file system (pmxcfs). | |
5771d9b0 TL |
169 | They are used to guarantee that each LRM is active once and working. As a |
170 | LRM only executes actions when it holds its lock we can mark a failed node | |
171 | as fenced if we can acquire its lock. This lets us then recover any failed | |
5eba0743 | 172 | HA services securely without any interference from the now unknown failed node. |
3810ae1e TL |
173 | This all gets supervised by the CRM which holds currently the manager master |
174 | lock. | |
175 | ||
c7470421 DM |
176 | |
177 | Service States | |
178 | ~~~~~~~~~~~~~~ | |
179 | ||
180 | The CRM use a service state enumeration to record the current service | |
181 | state. We display this state on the GUI and you can query it using | |
182 | the `ha-manager` command line tool: | |
183 | ||
184 | ---- | |
185 | # ha-manager status | |
186 | quorum OK | |
187 | master elsa (active, Mon Nov 21 07:23:29 2016) | |
188 | lrm elsa (active, Mon Nov 21 07:23:22 2016) | |
189 | service ct:100 (elsa, stopped) | |
190 | service ct:102 (elsa, started) | |
191 | service vm:501 (elsa, started) | |
192 | ---- | |
193 | ||
194 | Here is the list of possible states: | |
195 | ||
196 | stopped:: | |
197 | ||
198 | Service is stopped (confirmed by LRM). If the LRM detects a stopped | |
199 | service is still running, it will stop it again. | |
200 | ||
201 | request_stop:: | |
202 | ||
203 | Service should be stopped. The CRM waits for confirmation from the | |
204 | LRM. | |
205 | ||
1cd01666 DM |
206 | stopping:: |
207 | ||
208 | Pending stop request. But the CRM did not get the request so far. | |
209 | ||
c7470421 DM |
210 | started:: |
211 | ||
212 | Service is active an LRM should start it ASAP if not already running. | |
213 | If the Service fails and is detected to be not running the LRM | |
214 | restarts it | |
215 | (see xref:ha_manager_start_failure_policy[Start Failure Policy]). | |
216 | ||
1cd01666 DM |
217 | starting:: |
218 | ||
219 | Pending start request. But the CRM has not got any confirmation from the | |
220 | LRM that the service is running. | |
221 | ||
c7470421 DM |
222 | fence:: |
223 | ||
224 | Wait for node fencing (service node is not inside quorate cluster | |
225 | partition). As soon as node gets fenced successfully the service will | |
226 | be recovered to another node, if possible | |
227 | (see xref:ha_manager_fencing[Fencing]). | |
228 | ||
229 | freeze:: | |
230 | ||
231 | Do not touch the service state. We use this state while we reboot a | |
232 | node, or when we restart the LRM daemon | |
233 | (see xref:ha_manager_package_updates[Package Updates]). | |
234 | ||
235 | migrate:: | |
236 | ||
237 | Migrate service (live) to other node. | |
238 | ||
239 | error:: | |
240 | ||
241 | Service is disabled because of LRM errors. Needs manual intervention | |
242 | (see xref:ha_manager_error_recovery[Error Recovery]). | |
243 | ||
1cd01666 DM |
244 | queued:: |
245 | ||
246 | Service is newly added, and the CRM has not seen it so far. | |
247 | ||
248 | disabled:: | |
249 | ||
250 | Service is stopped and marked as `disabled` | |
251 | ||
c7470421 | 252 | |
3810ae1e TL |
253 | Local Resource Manager |
254 | ~~~~~~~~~~~~~~~~~~~~~~ | |
255 | ||
8c1189b6 | 256 | The local resource manager (`pve-ha-lrm`) is started as a daemon on |
3810ae1e TL |
257 | boot and waits until the HA cluster is quorate and thus cluster wide |
258 | locks are working. | |
259 | ||
260 | It can be in three states: | |
261 | ||
b8663359 | 262 | wait for agent lock:: |
e1ea726a FG |
263 | |
264 | The LRM waits for our exclusive lock. This is also used as idle state if no | |
265 | service is configured. | |
266 | ||
b8663359 | 267 | active:: |
e1ea726a FG |
268 | |
269 | The LRM holds its exclusive lock and has services configured. | |
270 | ||
b8663359 | 271 | lost agent lock:: |
e1ea726a FG |
272 | |
273 | The LRM lost its lock, this means a failure happened and quorum was lost. | |
3810ae1e TL |
274 | |
275 | After the LRM gets in the active state it reads the manager status | |
8c1189b6 | 276 | file in `/etc/pve/ha/manager_status` and determines the commands it |
2af6af05 | 277 | has to execute for the services it owns. |
3810ae1e | 278 | For each command a worker gets started, this workers are running in |
5eba0743 | 279 | parallel and are limited to at most 4 by default. This default setting |
8c1189b6 | 280 | may be changed through the datacenter configuration key `max_worker`. |
2af6af05 TL |
281 | When finished the worker process gets collected and its result saved for |
282 | the CRM. | |
3810ae1e | 283 | |
5eba0743 | 284 | .Maximum Concurrent Worker Adjustment Tips |
3810ae1e | 285 | [NOTE] |
5eba0743 | 286 | The default value of at most 4 concurrent workers may be unsuited for |
3810ae1e TL |
287 | a specific setup. For example may 4 live migrations happen at the same |
288 | time, which can lead to network congestions with slower networks and/or | |
289 | big (memory wise) services. Ensure that also in the worst case no congestion | |
8c1189b6 | 290 | happens and lower the `max_worker` value if needed. In the contrary, if you |
3810ae1e TL |
291 | have a particularly powerful high end setup you may also want to increase it. |
292 | ||
293 | Each command requested by the CRM is uniquely identifiable by an UID, when | |
294 | the worker finished its result will be processed and written in the LRM | |
8c1189b6 | 295 | status file `/etc/pve/nodes/<nodename>/lrm_status`. There the CRM may collect |
3810ae1e TL |
296 | it and let its state machine - respective the commands output - act on it. |
297 | ||
298 | The actions on each service between CRM and LRM are normally always synced. | |
299 | This means that the CRM requests a state uniquely marked by an UID, the LRM | |
300 | then executes this action *one time* and writes back the result, also | |
301 | identifiable by the same UID. This is needed so that the LRM does not | |
302 | executes an outdated command. | |
8c1189b6 | 303 | With the exception of the `stop` and the `error` command, |
c9aa5d47 | 304 | those two do not depend on the result produced and are executed |
3810ae1e TL |
305 | always in the case of the stopped state and once in the case of |
306 | the error state. | |
307 | ||
308 | .Read the Logs | |
309 | [NOTE] | |
310 | The HA Stack logs every action it makes. This helps to understand what | |
311 | and also why something happens in the cluster. Here its important to see | |
312 | what both daemons, the LRM and the CRM, did. You may use | |
313 | `journalctl -u pve-ha-lrm` on the node(s) where the service is and | |
314 | the same command for the pve-ha-crm on the node which is the current master. | |
315 | ||
316 | Cluster Resource Manager | |
317 | ~~~~~~~~~~~~~~~~~~~~~~~~ | |
22653ac8 | 318 | |
8c1189b6 | 319 | The cluster resource manager (`pve-ha-crm`) starts on each node and |
22653ac8 DM |
320 | waits there for the manager lock, which can only be held by one node |
321 | at a time. The node which successfully acquires the manager lock gets | |
3810ae1e TL |
322 | promoted to the CRM master. |
323 | ||
2af6af05 | 324 | It can be in three states: |
3810ae1e | 325 | |
b8663359 | 326 | wait for agent lock:: |
e1ea726a | 327 | |
97ae300a | 328 | The CRM waits for our exclusive lock. This is also used as idle state if no |
e1ea726a FG |
329 | service is configured |
330 | ||
b8663359 | 331 | active:: |
e1ea726a | 332 | |
97ae300a | 333 | The CRM holds its exclusive lock and has services configured |
e1ea726a | 334 | |
b8663359 | 335 | lost agent lock:: |
e1ea726a | 336 | |
97ae300a | 337 | The CRM lost its lock, this means a failure happened and quorum was lost. |
3810ae1e TL |
338 | |
339 | It main task is to manage the services which are configured to be highly | |
4c34defd TL |
340 | available and try to always enforce the requested state. For example, a |
341 | service with the requested state 'started' will be started if its not | |
342 | already running. If it crashes it will be automatically started again. | |
343 | Thus the CRM dictates the actions which the LRM needs to execute. | |
22653ac8 DM |
344 | |
345 | When an node leaves the cluster quorum, its state changes to unknown. | |
346 | If the current CRM then can secure the failed nodes lock, the services | |
347 | will be 'stolen' and restarted on another node. | |
348 | ||
349 | When a cluster member determines that it is no longer in the cluster | |
350 | quorum, the LRM waits for a new quorum to form. As long as there is no | |
351 | quorum the node cannot reset the watchdog. This will trigger a reboot | |
2af6af05 | 352 | after the watchdog then times out, this happens after 60 seconds. |
22653ac8 | 353 | |
85363588 | 354 | |
2b52e195 | 355 | Configuration |
22653ac8 DM |
356 | ------------- |
357 | ||
85363588 DM |
358 | The HA stack is well integrated into the {pve} API. So, for example, |
359 | HA can be configured via the `ha-manager` command line interface, or | |
360 | the {pve} web interface - both interfaces provide an easy way to | |
361 | manage HA. Automation tools can use the API directly. | |
362 | ||
363 | All HA configuration files are within `/etc/pve/ha/`, so they get | |
364 | automatically distributed to the cluster nodes, and all nodes share | |
365 | the same HA configuration. | |
366 | ||
206c2476 | 367 | |
4c34defd | 368 | [[ha_manager_resource_config]] |
206c2476 DM |
369 | Resources |
370 | ~~~~~~~~~ | |
371 | ||
863a8f3a DM |
372 | [thumbnail="gui-ha-manager-status.png"] |
373 | ||
4d63b3cc | 374 | |
85363588 DM |
375 | The resource configuration file `/etc/pve/ha/resources.cfg` stores |
376 | the list of resources managed by `ha-manager`. A resource configuration | |
377 | inside that list look like this: | |
378 | ||
379 | ---- | |
8bdc398c | 380 | <type>: <name> |
85363588 DM |
381 | <property> <value> |
382 | ... | |
383 | ---- | |
384 | ||
698e5dd2 DM |
385 | It starts with a resource type followed by a resource specific name, |
386 | separated with colon. Together this forms the HA resource ID, which is | |
387 | used by all `ha-manager` commands to uniquely identify a resource | |
a9c77fec DM |
388 | (example: `vm:100` or `ct:101`). The next lines contain additional |
389 | properties: | |
85363588 DM |
390 | |
391 | include::ha-resources-opts.adoc[] | |
392 | ||
8bdc398c DM |
393 | Here is a real world example with one VM and one container. As you see, |
394 | the syntax of those files is really simple, so it is even posiible to | |
395 | read or edit those files using your favorite editor: | |
396 | ||
e7b9b0ac | 397 | .Configuration Example (`/etc/pve/ha/resources.cfg`) |
8bdc398c DM |
398 | ---- |
399 | vm: 501 | |
400 | state started | |
401 | max_relocate 2 | |
402 | ||
403 | ct: 102 | |
a319e18b DM |
404 | # Note: use default settings for everything |
405 | ---- | |
406 | ||
4d63b3cc DM |
407 | [thumbnail="gui-ha-manager-add-resource.png"] |
408 | ||
a319e18b DM |
409 | Above config was generated using the `ha-manager` command line tool: |
410 | ||
411 | ---- | |
412 | # ha-manager add vm:501 --state started --max_relocate 2 | |
413 | # ha-manager add ct:102 | |
8bdc398c DM |
414 | ---- |
415 | ||
85363588 | 416 | |
1acab952 | 417 | [[ha_manager_groups]] |
206c2476 DM |
418 | Groups |
419 | ~~~~~~ | |
420 | ||
4d63b3cc DM |
421 | [thumbnail="gui-ha-manager-groups-view.png"] |
422 | ||
85363588 DM |
423 | The HA group configuration file `/etc/pve/ha/groups.cfg` is used to |
424 | define groups of cluster nodes. A resource can be restricted to run | |
206c2476 DM |
425 | only on the members of such group. A group configuration look like |
426 | this: | |
85363588 | 427 | |
206c2476 DM |
428 | ---- |
429 | group: <group> | |
430 | nodes <node_list> | |
431 | <property> <value> | |
432 | ... | |
433 | ---- | |
85363588 | 434 | |
206c2476 | 435 | include::ha-groups-opts.adoc[] |
22653ac8 | 436 | |
4d63b3cc DM |
437 | [thumbnail="gui-ha-manager-add-group.png"] |
438 | ||
1acab952 DM |
439 | A commom requirement is that a resource should run on a specific |
440 | node. Usually the resource is able to run on other nodes, so you can define | |
441 | an unrestricted group with a single member: | |
442 | ||
443 | ---- | |
444 | # ha-manager groupadd prefer_node1 --nodes node1 | |
445 | ---- | |
446 | ||
447 | For bigger clusters, it makes sense to define a more detailed failover | |
448 | behavior. For example, you may want to run a set of services on | |
449 | `node1` if possible. If `node1` is not available, you want to run them | |
450 | equally splitted on `node2` and `node3`. If those nodes also fail the | |
451 | services should run on `node4`. To achieve this you could set the node | |
452 | list to: | |
453 | ||
454 | ---- | |
455 | # ha-manager groupadd mygroup1 -nodes "node1:2,node2:1,node3:1,node4" | |
456 | ---- | |
457 | ||
458 | Another use case is if a resource uses other resources only available | |
459 | on specific nodes, lets say `node1` and `node2`. We need to make sure | |
460 | that HA manager does not use other nodes, so we need to create a | |
461 | restricted group with said nodes: | |
462 | ||
463 | ---- | |
464 | # ha-manager groupadd mygroup2 -nodes "node1,node2" -restricted | |
465 | ---- | |
466 | ||
467 | Above commands created the following group configuration fils: | |
468 | ||
469 | .Configuration Example (`/etc/pve/ha/groups.cfg`) | |
470 | ---- | |
471 | group: prefer_node1 | |
472 | nodes node1 | |
473 | ||
474 | group: mygroup1 | |
475 | nodes node2:1,node4,node1:2,node3:1 | |
476 | ||
477 | group: mygroup2 | |
478 | nodes node2,node1 | |
479 | restricted 1 | |
480 | ---- | |
481 | ||
482 | ||
483 | The `nofailback` options is mostly useful to avoid unwanted resource | |
484 | movements during administartion tasks. For example, if you need to | |
485 | migrate a service to a node which hasn't the highest priority in the | |
486 | group, you need to tell the HA manager to not move this service | |
487 | instantly back by setting the `nofailback` option. | |
488 | ||
489 | Another scenario is when a service was fenced and it got recovered to | |
490 | another node. The admin tries to repair the fenced node and brings it | |
491 | up online again to investigate the failure cause and check if it runs | |
492 | stable again. Setting the `nofailback` flag prevents that the | |
493 | recovered services move straight back to the fenced node. | |
494 | ||
22653ac8 | 495 | |
80c0adcb | 496 | [[ha_manager_fencing]] |
3810ae1e TL |
497 | Fencing |
498 | ------- | |
499 | ||
0d427077 DM |
500 | On node failures, fencing ensures that the erroneous node is |
501 | guaranteed to be offline. This is required to make sure that no | |
502 | resource runs twice when it gets recovered on another node. This is a | |
503 | really important task, because without, it would not be possible to | |
504 | recover a resource on another node. | |
505 | ||
506 | If a node would not get fenced, it would be in an unknown state where | |
507 | it may have still access to shared resources. This is really | |
508 | dangerous! Imagine that every network but the storage one broke. Now, | |
509 | while not reachable from the public network, the VM still runs and | |
510 | writes to the shared storage. | |
511 | ||
512 | If we then simply start up this VM on another node, we would get a | |
513 | dangerous race conditions because we write from both nodes. Such | |
514 | condition can destroy all VM data and the whole VM could be rendered | |
515 | unusable. The recovery could also fail if the storage protects from | |
516 | multiple mounts. | |
517 | ||
5771d9b0 TL |
518 | |
519 | How {pve} Fences | |
0d427077 | 520 | ~~~~~~~~~~~~~~~~ |
5771d9b0 | 521 | |
61972f55 DM |
522 | There are different methods to fence a node, for example, fence |
523 | devices which cut off the power from the node or disable their | |
524 | communication completely. Those are often quite expensive and bring | |
525 | additional critical components into a system, because if they fail you | |
526 | cannot recover any service. | |
527 | ||
528 | We thus wanted to integrate a simpler fencing method, which does not | |
529 | require additional external hardware. This can be done using | |
530 | watchdog timers. | |
531 | ||
532 | .Possible Fencing Methods | |
533 | - external power switches | |
534 | - isolate nodes by disabling complete network traffic on the switch | |
535 | - self fencing using watchdog timers | |
536 | ||
537 | Watchdog timers are widely used in critical and dependable systems | |
538 | since the beginning of micro controllers. They are often independent | |
539 | and simple integrated circuits which are used to detect and recover | |
540 | from computer malfunctions. | |
541 | ||
542 | During normal operation, `ha-manager` regularly resets the watchdog | |
543 | timer to prevent it from elapsing. If, due to a hardware fault or | |
544 | program error, the computer fails to reset the watchdog, the timer | |
545 | will elapse and triggers a reset of the whole server (reboot). | |
546 | ||
547 | Recent server motherboards often include such hardware watchdogs, but | |
548 | these need to be configured. If no watchdog is available or | |
549 | configured, we fall back to the Linux Kernel 'softdog'. While still | |
550 | reliable, it is not independent of the servers hardware, and thus has | |
551 | a lower reliability than a hardware watchdog. | |
3810ae1e | 552 | |
a472fde8 | 553 | |
3810ae1e TL |
554 | Configure Hardware Watchdog |
555 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
a472fde8 DM |
556 | |
557 | By default, all hardware watchdog modules are blocked for security | |
558 | reasons. They are like a loaded gun if not correctly initialized. To | |
559 | enable a hardware watchdog, you need to specify the module to load in | |
560 | '/etc/default/pve-ha-manager', for example: | |
561 | ||
562 | ---- | |
563 | # select watchdog module (default is softdog) | |
564 | WATCHDOG_MODULE=iTCO_wdt | |
565 | ---- | |
566 | ||
567 | This configuration is read by the 'watchdog-mux' service, which load | |
568 | the specified module at startup. | |
569 | ||
3810ae1e | 570 | |
2957ef80 TL |
571 | Recover Fenced Services |
572 | ~~~~~~~~~~~~~~~~~~~~~~~ | |
573 | ||
480e67e1 DM |
574 | After a node failed and its fencing was successful, the CRM tries to |
575 | move services from the failed node to nodes which are still online. | |
576 | ||
577 | The selection of nodes, on which those services gets recovered, is | |
578 | influenced by the resource `group` settings, the list of currently active | |
579 | nodes, and their respective active service count. | |
580 | ||
581 | The CRM first builds a set out of the intersection between user selected | |
582 | nodes (from `group` setting) and available nodes. It then choose the | |
583 | subset of nodes with the highest priority, and finally select the node | |
584 | with the lowest active service count. This minimizes the possibility | |
585 | of an overloaded node. | |
586 | ||
587 | CAUTION: On node failure, the CRM distributes services to the | |
588 | remaining nodes. This increase the service count on those nodes, and | |
589 | can lead to high load, especially on small clusters. Please design | |
590 | your cluster so that it can handle such worst case scenarios. | |
2957ef80 | 591 | |
22653ac8 | 592 | |
c7470421 | 593 | [[ha_manager_start_failure_policy]] |
a3189ad1 TL |
594 | Start Failure Policy |
595 | --------------------- | |
596 | ||
597 | The start failure policy comes in effect if a service failed to start on a | |
598 | node once ore more times. It can be used to configure how often a restart | |
599 | should be triggered on the same node and how often a service should be | |
600 | relocated so that it gets a try to be started on another node. | |
601 | The aim of this policy is to circumvent temporary unavailability of shared | |
602 | resources on a specific node. For example, if a shared storage isn't available | |
603 | on a quorate node anymore, e.g. network problems, but still on other nodes, | |
604 | the relocate policy allows then that the service gets started nonetheless. | |
605 | ||
606 | There are two service start recover policy settings which can be configured | |
22653ac8 DM |
607 | specific for each resource. |
608 | ||
609 | max_restart:: | |
610 | ||
5eba0743 | 611 | Maximum number of tries to restart an failed service on the actual |
22653ac8 DM |
612 | node. The default is set to one. |
613 | ||
614 | max_relocate:: | |
615 | ||
5eba0743 | 616 | Maximum number of tries to relocate the service to a different node. |
22653ac8 DM |
617 | A relocate only happens after the max_restart value is exceeded on the |
618 | actual node. The default is set to one. | |
619 | ||
0abc65b0 | 620 | NOTE: The relocate count state will only reset to zero when the |
22653ac8 | 621 | service had at least one successful start. That means if a service is |
4c34defd | 622 | re-started without fixing the error only the restart policy gets |
22653ac8 DM |
623 | repeated. |
624 | ||
c7470421 DM |
625 | |
626 | [[ha_manager_error_recovery]] | |
2b52e195 | 627 | Error Recovery |
22653ac8 DM |
628 | -------------- |
629 | ||
630 | If after all tries the service state could not be recovered it gets | |
631 | placed in an error state. In this state the service won't get touched | |
c5bca1ae | 632 | by the HA stack anymore. The only way out is disabling a service: |
d02982f7 | 633 | |
c5bca1ae TL |
634 | ---- |
635 | # ha-manager set vm:100 --state disabled | |
636 | ---- | |
d02982f7 | 637 | |
c5bca1ae TL |
638 | This can also be done in the web interface. |
639 | ||
640 | To recover from the error state you should do the following: | |
22653ac8 | 641 | |
c5bca1ae TL |
642 | * bring the resource back into a safe and consistent state (e.g.: |
643 | kill its process if the service could not be stopped) | |
22653ac8 | 644 | |
c5bca1ae | 645 | * disable the resource to remove the error flag |
22653ac8 DM |
646 | |
647 | * fix the error which led to this failures | |
648 | ||
4c34defd | 649 | * *after* you fixed all errors you may request that the service starts again |
22653ac8 DM |
650 | |
651 | ||
26513dae DM |
652 | [[ha_manager_package_updates]] |
653 | Package Updates | |
654 | --------------- | |
655 | ||
656 | When updating the ha-manager you should do one node after the other, never | |
657 | all at once for various reasons. First, while we test our software | |
658 | thoughtfully, a bug affecting your specific setup cannot totally be ruled out. | |
659 | Upgrading one node after the other and checking the functionality of each node | |
660 | after finishing the update helps to recover from an eventual problems, while | |
661 | updating all could render you in a broken cluster state and is generally not | |
662 | good practice. | |
663 | ||
664 | Also, the {pve} HA stack uses a request acknowledge protocol to perform | |
665 | actions between the cluster and the local resource manager. For restarting, | |
666 | the LRM makes a request to the CRM to freeze all its services. This prevents | |
667 | that they get touched by the Cluster during the short time the LRM is restarting. | |
668 | After that the LRM may safely close the watchdog during a restart. | |
669 | Such a restart happens on a update and as already stated a active master | |
670 | CRM is needed to acknowledge the requests from the LRM, if this is not the case | |
671 | the update process can be too long which, in the worst case, may result in | |
672 | a watchdog reset. | |
673 | ||
674 | ||
a9023144 DM |
675 | Node Maintenance |
676 | ---------------- | |
52a75187 | 677 | |
a9023144 DM |
678 | It is sometimes possible to shutdown or reboot a node to do |
679 | maintenance tasks. Either to replace hardware, or simply to install a | |
680 | new kernel image. | |
681 | ||
682 | ||
683 | Shutdown | |
684 | ~~~~~~~~ | |
685 | ||
686 | A shutdown ('poweroff') is usually done if the node is planned to stay | |
687 | down for some time. The LRM stops all managed services in that | |
688 | case. This means that other nodes will take over those service | |
689 | afterwards. | |
690 | ||
691 | NOTE: Recent hardware has large amounts of RAM. So we stop all | |
692 | resources, then restart them to avoid online migration of all that | |
693 | RAM. If you want to use online migration, you need to invoke that | |
694 | manually before you shutdown the node. | |
695 | ||
696 | ||
697 | Reboot | |
698 | ~~~~~~ | |
699 | ||
700 | Node reboots are initiated with the 'reboot' command. This is usually | |
701 | done after installing a new kernel. Please note that this is different | |
702 | from ``shutdown'', because the node immediately starts again. | |
703 | ||
704 | The LRM tells the CRM that it wants to restart, and waits until the | |
26513dae DM |
705 | CRM puts all resources into the `freeze` state (same mechanism is used |
706 | for xref:ha_manager_package_updates[Pakage Updates]). This prevents | |
707 | that those resources are moved to other nodes. Instead, the CRM start | |
708 | the resources after the reboot on the same node. | |
a9023144 DM |
709 | |
710 | ||
711 | Manual Resource Movement | |
712 | ~~~~~~~~~~~~~~~~~~~~~~~~ | |
713 | ||
714 | Last but not least, you can also move resources manually to other | |
715 | nodes before you shutdown or restart a node. The advantage is that you | |
716 | have full control, and you can decide if you want to use online | |
717 | migration or not. | |
718 | ||
719 | NOTE: Please do not 'kill' services like `pve-ha-crm`, `pve-ha-lrm` or | |
720 | `watchdog-mux`. They manage and use the watchdog, so this can result | |
721 | in a node reboot. | |
52a75187 DM |
722 | |
723 | ||
8b598c33 | 724 | [[ha_manager_service_operations]] |
2b52e195 | 725 | Service Operations |
22653ac8 DM |
726 | ------------------ |
727 | ||
728 | This are how the basic user-initiated service operations (via | |
8c1189b6 | 729 | `ha-manager`) work. |
22653ac8 | 730 | |
4c34defd | 731 | set state:: |
22653ac8 | 732 | |
4c34defd TL |
733 | Request the service state. |
734 | See xref:ha_manager_resource_config[Resource Configuration] for possible | |
735 | request states. | |
c402b37f | 736 | + |
4c34defd TL |
737 | ---- |
738 | # ha-manager set SID -state REQUEST_STATE | |
739 | ---- | |
22653ac8 DM |
740 | |
741 | disable:: | |
742 | ||
4c34defd TL |
743 | The service will be placed in the stopped state, even if it was in the error |
744 | state. The service will not be recovered on a node failure and will stay | |
745 | stopped while it is in this state. | |
22653ac8 DM |
746 | |
747 | migrate/relocate:: | |
748 | ||
5eba0743 | 749 | The service will be relocated (live) to another node. |
22653ac8 DM |
750 | |
751 | remove:: | |
752 | ||
5eba0743 | 753 | The service will be removed from the HA managed resource list. Its |
22653ac8 DM |
754 | current state will not be touched. |
755 | ||
756 | start/stop:: | |
757 | ||
8c1189b6 FG |
758 | `start` and `stop` commands can be issued to the resource specific tools |
759 | (like `qm` or `pct`), they will forward the request to the | |
760 | `ha-manager` which then will execute the action and set the resulting | |
22653ac8 DM |
761 | service state (enabled, disabled). |
762 | ||
763 | ||
22653ac8 DM |
764 | ifdef::manvolnum[] |
765 | include::pve-copyright.adoc[] | |
766 | endif::manvolnum[] | |
767 |