5 include::attributes.txt[]
11 ha-manager - Proxmox VE HA Manager
16 include::ha-manager.1-synopsis.adoc[]
24 include::attributes.txt[]
28 Our modern society depends heavily on information provided by
29 computers over the network. Mobile devices amplified that dependency,
30 because people can access the network any time from anywhere. If you
31 provide such services, it is very important that they are available
34 We can mathematically define the availability as the ratio of (A) the
35 total time a service is capable of being used during a given interval
36 to (B) the length of the interval. It is normally expressed as a
37 percentage of uptime in a given year.
39 .Availability - Downtime per Year
40 [width="60%",cols="<d,d",options="header"]
41 |===========================================================
42 |Availability % |Downtime per year
47 |99.9999 |31.5 seconds
48 |99.99999 |3.15 seconds
49 |===========================================================
51 There are several ways to increase availability. The most elegant
52 solution is to rewrite your software, so that you can run it on
53 several host at the same time. The software itself need to have a way
54 to detect errors and do failover. This is relatively easy if you just
55 want to serve read-only web pages. But in general this is complex, and
56 sometimes impossible because you cannot modify the software
57 yourself. The following solutions works without modifying the
60 * Use reliable ``server'' components
62 NOTE: Computer components with same functionality can have varying
63 reliability numbers, depending on the component quality. Most vendors
64 sell components with higher reliability as ``server'' components -
65 usually at higher price.
67 * Eliminate single point of failure (redundant components)
68 ** use an uninterruptible power supply (UPS)
69 ** use redundant power supplies on the main boards
71 ** use redundant network hardware
72 ** use RAID for local storage
73 ** use distributed, redundant storage for VM data
76 ** rapidly accessible administrators (24/7)
77 ** availability of spare parts (other nodes in a {pve} cluster)
78 ** automatic error detection (provided by `ha-manager`)
79 ** automatic failover (provided by `ha-manager`)
81 Virtualization environments like {pve} make it much easier to reach
82 high availability because they remove the ``hardware'' dependency. They
83 also support to setup and use redundant storage and network
84 devices. So if one host fail, you can simply start those services on
85 another host within your cluster.
87 Even better, {pve} provides a software stack called `ha-manager`,
88 which can do that automatically for you. It is able to automatically
89 detect errors and do automatic failover.
91 {pve} `ha-manager` works like an ``automated'' administrator. First, you
92 configure what resources (VMs, containers, ...) it should
93 manage. `ha-manager` then observes correct functionality, and handles
94 service failover to another node in case of errors. `ha-manager` can
95 also handle normal user requests which may start, stop, relocate and
98 But high availability comes at a price. High quality components are
99 more expensive, and making them redundant duplicates the costs at
100 least. Additional spare parts increase costs further. So you should
101 carefully calculate the benefits, and compare with those additional
104 TIP: Increasing availability from 99% to 99.9% is relatively
105 simply. But increasing availability from 99.9999% to 99.99999% is very
106 hard and costly. `ha-manager` has typical error detection and failover
107 times of about 2 minutes, so you can get no more than 99.999%
113 * at least three cluster nodes (to get reliable quorum)
115 * shared storage for VMs and containers
117 * hardware redundancy (everywhere)
119 * hardware watchdog - if not available we fall back to the
120 linux kernel software watchdog (`softdog`)
122 * optional hardware fencing devices
125 [[ha_manager_resources]]
129 We call the primary management unit handled by `ha-manager` a
130 resource. A resource (also called ``service'') is uniquely
131 identified by a service ID (SID), which consists of the resource type
132 and an type specific ID, e.g.: `vm:100`. That example would be a
133 resource of type `vm` (virtual machine) with the ID 100.
135 For now we have two important resources types - virtual machines and
136 containers. One basic idea here is that we can bundle related software
137 into such VM or container, so there is no need to compose one big
138 service from other services, like it was done with `rgmanager`. In
139 general, a HA enabled resource should not depend on other resources.
145 This section provides an in detail description of the {PVE} HA-manager
146 internals. It describes how the CRM and the LRM work together.
148 To provide High Availability two daemons run on each node:
152 The local resource manager (LRM), it controls the services running on
154 It reads the requested states for its services from the current manager
155 status file and executes the respective commands.
159 The cluster resource manager (CRM), it controls the cluster wide
160 actions of the services, processes the LRM results and includes the state
161 machine which controls the state of each service.
163 .Locks in the LRM & CRM
165 Locks are provided by our distributed configuration file system (pmxcfs).
166 They are used to guarantee that each LRM is active once and working. As a
167 LRM only executes actions when it holds its lock we can mark a failed node
168 as fenced if we can acquire its lock. This lets us then recover any failed
169 HA services securely without any interference from the now unknown failed node.
170 This all gets supervised by the CRM which holds currently the manager master
173 Local Resource Manager
174 ~~~~~~~~~~~~~~~~~~~~~~
176 The local resource manager (`pve-ha-lrm`) is started as a daemon on
177 boot and waits until the HA cluster is quorate and thus cluster wide
180 It can be in three states:
182 wait for agent lock::
184 The LRM waits for our exclusive lock. This is also used as idle state if no
185 service is configured.
189 The LRM holds its exclusive lock and has services configured.
193 The LRM lost its lock, this means a failure happened and quorum was lost.
195 After the LRM gets in the active state it reads the manager status
196 file in `/etc/pve/ha/manager_status` and determines the commands it
197 has to execute for the services it owns.
198 For each command a worker gets started, this workers are running in
199 parallel and are limited to at most 4 by default. This default setting
200 may be changed through the datacenter configuration key `max_worker`.
201 When finished the worker process gets collected and its result saved for
204 .Maximum Concurrent Worker Adjustment Tips
206 The default value of at most 4 concurrent workers may be unsuited for
207 a specific setup. For example may 4 live migrations happen at the same
208 time, which can lead to network congestions with slower networks and/or
209 big (memory wise) services. Ensure that also in the worst case no congestion
210 happens and lower the `max_worker` value if needed. In the contrary, if you
211 have a particularly powerful high end setup you may also want to increase it.
213 Each command requested by the CRM is uniquely identifiable by an UID, when
214 the worker finished its result will be processed and written in the LRM
215 status file `/etc/pve/nodes/<nodename>/lrm_status`. There the CRM may collect
216 it and let its state machine - respective the commands output - act on it.
218 The actions on each service between CRM and LRM are normally always synced.
219 This means that the CRM requests a state uniquely marked by an UID, the LRM
220 then executes this action *one time* and writes back the result, also
221 identifiable by the same UID. This is needed so that the LRM does not
222 executes an outdated command.
223 With the exception of the `stop` and the `error` command,
224 those two do not depend on the result produced and are executed
225 always in the case of the stopped state and once in the case of
230 The HA Stack logs every action it makes. This helps to understand what
231 and also why something happens in the cluster. Here its important to see
232 what both daemons, the LRM and the CRM, did. You may use
233 `journalctl -u pve-ha-lrm` on the node(s) where the service is and
234 the same command for the pve-ha-crm on the node which is the current master.
236 Cluster Resource Manager
237 ~~~~~~~~~~~~~~~~~~~~~~~~
239 The cluster resource manager (`pve-ha-crm`) starts on each node and
240 waits there for the manager lock, which can only be held by one node
241 at a time. The node which successfully acquires the manager lock gets
242 promoted to the CRM master.
244 It can be in three states:
246 wait for agent lock::
248 The CRM waits for our exclusive lock. This is also used as idle state if no
249 service is configured
253 The CRM holds its exclusive lock and has services configured
257 The CRM lost its lock, this means a failure happened and quorum was lost.
259 It main task is to manage the services which are configured to be highly
260 available and try to always enforce them to the wanted state, e.g.: a
261 enabled service will be started if its not running, if it crashes it will
262 be started again. Thus it dictates the LRM the actions it needs to execute.
264 When an node leaves the cluster quorum, its state changes to unknown.
265 If the current CRM then can secure the failed nodes lock, the services
266 will be 'stolen' and restarted on another node.
268 When a cluster member determines that it is no longer in the cluster
269 quorum, the LRM waits for a new quorum to form. As long as there is no
270 quorum the node cannot reset the watchdog. This will trigger a reboot
271 after the watchdog then times out, this happens after 60 seconds.
276 The HA stack is well integrated in the Proxmox VE API2. So, for
277 example, HA can be configured via `ha-manager` or the PVE web
278 interface, which both provide an easy to use tool.
280 The resource configuration file can be located at
281 `/etc/pve/ha/resources.cfg` and the group configuration file at
282 `/etc/pve/ha/groups.cfg`. Use the provided tools to make changes,
283 there shouldn't be any need to edit them manually.
288 If a node needs maintenance you should migrate and or relocate all
289 services which are required to run always on another node first.
290 After that you can stop the LRM and CRM services. But note that the
291 watchdog triggers if you stop it with active services.
296 When updating the ha-manager you should do one node after the other, never
297 all at once for various reasons. First, while we test our software
298 thoughtfully, a bug affecting your specific setup cannot totally be ruled out.
299 Upgrading one node after the other and checking the functionality of each node
300 after finishing the update helps to recover from an eventual problems, while
301 updating all could render you in a broken cluster state and is generally not
304 Also, the {pve} HA stack uses a request acknowledge protocol to perform
305 actions between the cluster and the local resource manager. For restarting,
306 the LRM makes a request to the CRM to freeze all its services. This prevents
307 that they get touched by the Cluster during the short time the LRM is restarting.
308 After that the LRM may safely close the watchdog during a restart.
309 Such a restart happens on a update and as already stated a active master
310 CRM is needed to acknowledge the requests from the LRM, if this is not the case
311 the update process can be too long which, in the worst case, may result in
315 [[ha_manager_fencing]]
322 Fencing secures that on a node failure the dangerous node gets will be rendered
323 unable to do any damage and that no resource runs twice when it gets recovered
324 from the failed node. This is a really important task and one of the base
325 principles to make a system Highly Available.
327 If a node would not get fenced it would be in an unknown state where it may
328 have still access to shared resources, this is really dangerous!
329 Imagine that every network but the storage one broke, now while not
330 reachable from the public network the VM still runs and writes on the shared
331 storage. If we would not fence the node and just start up this VM on another
332 Node we would get dangerous race conditions, atomicity violations the whole VM
333 could be rendered unusable. The recovery could also simply fail if the storage
334 protects from multiple mounts and thus defeat the purpose of HA.
339 There are different methods to fence a node, for example fence devices which
340 cut off the power from the node or disable their communication completely.
342 Those are often quite expensive and bring additional critical components in
343 a system, because if they fail you cannot recover any service.
345 We thus wanted to integrate a simpler method in the HA Manager first, namely
346 self fencing with watchdogs.
348 Watchdogs are widely used in critical and dependable systems since the
349 beginning of micro controllers, they are often independent and simple
350 integrated circuit which programs can use to watch them. After opening they need to
351 report periodically. If, for whatever reason, a program becomes unable to do
352 so the watchdogs triggers a reset of the whole server.
354 Server motherboards often already include such hardware watchdogs, these need
355 to be configured. If no watchdog is available or configured we fall back to the
356 Linux Kernel softdog while still reliable it is not independent of the servers
357 Hardware and thus has a lower reliability then a hardware watchdog.
359 Configure Hardware Watchdog
360 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
361 By default all watchdog modules are blocked for security reasons as they are
362 like a loaded gun if not correctly initialized.
363 If you have a hardware watchdog available remove its kernel module from the
364 blacklist, load it with insmod and restart the `watchdog-mux` service or reboot
367 Recover Fenced Services
368 ~~~~~~~~~~~~~~~~~~~~~~~
370 After a node failed and its fencing was successful we start to recover services
371 to other available nodes and restart them there so that they can provide service
374 The selection of the node on which the services gets recovered is influenced
375 by the users group settings, the currently active nodes and their respective
376 active service count.
377 First we build a set out of the intersection between user selected nodes and
378 available nodes. Then the subset with the highest priority of those nodes
379 gets chosen as possible nodes for recovery. We select the node with the
380 currently lowest active service count as a new node for the service.
381 That minimizes the possibility of an overload, which else could cause an
382 unresponsive node and as a result a chain reaction of node failures in the
385 [[ha_manager_groups]]
389 A group is a collection of cluster nodes which a service may be bound to.
396 List of group node members where a priority can be given to each node.
397 A service bound to this group will run on the nodes with the highest priority
398 available. If more nodes are in the highest priority class the services will
399 get distributed to those node if not already there. The priorities have a
400 relative meaning only.
402 You want to run all services from a group on `node1` if possible. If this node
403 is not available, you want them to run equally splitted on `node2` and `node3`, and
404 if those fail it should use `node4`.
405 To achieve this you could set the node list to:
407 ha-manager groupset mygroup -nodes "node1:2,node2:1,node3:1,node4"
411 Resources bound to this group may only run on nodes defined by the
412 group. If no group node member is available the resource will be
413 placed in the stopped state.
415 Lets say a service uses resources only available on `node1` and `node2`,
416 so we need to make sure that HA manager does not use other nodes.
417 We need to create a 'restricted' group with said nodes:
419 ha-manager groupset mygroup -nodes "node1,node2" -restricted
423 The resource won't automatically fail back when a more preferred node
424 (re)joins the cluster.
426 * You need to migrate a service to a node which hasn't the highest priority
427 in the group at the moment, to tell the HA manager to not move this service
428 instantly back set the 'nofailback' option and the service will stay on
431 * A service was fenced and it got recovered to another node. The admin
432 repaired the node and brought it up online again but does not want that the
433 recovered services move straight back to the repaired node as he wants to
434 first investigate the failure cause and check if it runs stable. He can use
435 the 'nofailback' option to achieve this.
439 ---------------------
441 The start failure policy comes in effect if a service failed to start on a
442 node once ore more times. It can be used to configure how often a restart
443 should be triggered on the same node and how often a service should be
444 relocated so that it gets a try to be started on another node.
445 The aim of this policy is to circumvent temporary unavailability of shared
446 resources on a specific node. For example, if a shared storage isn't available
447 on a quorate node anymore, e.g. network problems, but still on other nodes,
448 the relocate policy allows then that the service gets started nonetheless.
450 There are two service start recover policy settings which can be configured
451 specific for each resource.
455 Maximum number of tries to restart an failed service on the actual
456 node. The default is set to one.
460 Maximum number of tries to relocate the service to a different node.
461 A relocate only happens after the max_restart value is exceeded on the
462 actual node. The default is set to one.
464 NOTE: The relocate count state will only reset to zero when the
465 service had at least one successful start. That means if a service is
466 re-enabled without fixing the error only the restart policy gets
472 If after all tries the service state could not be recovered it gets
473 placed in an error state. In this state the service won't get touched
474 by the HA stack anymore. To recover from this state you should follow
477 * bring the resource back into a safe and consistent state (e.g.,
480 * disable the ha resource to place it in an stopped state
482 * fix the error which led to this failures
484 * *after* you fixed all errors you may enable the service again
487 [[ha_manager_service_operations]]
491 This are how the basic user-initiated service operations (via
496 The service will be started by the LRM if not already running.
500 The service will be stopped by the LRM if running.
504 The service will be relocated (live) to another node.
508 The service will be removed from the HA managed resource list. Its
509 current state will not be touched.
513 `start` and `stop` commands can be issued to the resource specific tools
514 (like `qm` or `pct`), they will forward the request to the
515 `ha-manager` which then will execute the action and set the resulting
516 service state (enabled, disabled).
524 Service is stopped (confirmed by LRM), if detected running it will get stopped
529 Service should be stopped. Waiting for confirmation from LRM.
533 Service is active an LRM should start it ASAP if not already running.
534 If the Service fails and is detected to be not running the LRM restarts it.
538 Wait for node fencing (service node is not inside quorate cluster
540 As soon as node gets fenced successfully the service will be recovered to
541 another node, if possible.
545 Do not touch the service state. We use this state while we reboot a
546 node, or when we restart the LRM daemon.
550 Migrate service (live) to other node.
554 Service disabled because of LRM errors. Needs manual intervention.
558 include::pve-copyright.adoc[]