asciidoc.js: use mediawiki mw.hook
[pve-docs.git] / qm.adoc
CommitLineData
80c0adcb 1[[chapter_virtual_machines]]
f69cfd23 2ifdef::manvolnum[]
b2f242ab
DM
3qm(1)
4=====
38fd0958 5include::attributes.txt[]
5f09af76
DM
6:pve-toplevel:
7
f69cfd23
DM
8NAME
9----
10
11qm - Qemu/KVM Virtual Machine Manager
12
13
49a5e11c 14SYNOPSIS
f69cfd23
DM
15--------
16
17include::qm.1-synopsis.adoc[]
18
19DESCRIPTION
20-----------
21endif::manvolnum[]
f69cfd23
DM
22ifndef::manvolnum[]
23Qemu/KVM Virtual Machines
24=========================
38fd0958 25include::attributes.txt[]
5f09af76 26:pve-toplevel:
194d2f29 27endif::manvolnum[]
5f09af76 28
c4cba5d7
EK
29// deprecates
30// http://pve.proxmox.com/wiki/Container_and_Full_Virtualization
31// http://pve.proxmox.com/wiki/KVM
32// http://pve.proxmox.com/wiki/Qemu_Server
33
5eba0743 34Qemu (short form for Quick Emulator) is an open source hypervisor that emulates a
c4cba5d7
EK
35physical computer. From the perspective of the host system where Qemu is
36running, Qemu is a user program which has access to a number of local resources
37like partitions, files, network cards which are then passed to an
189d3661 38emulated computer which sees them as if they were real devices.
c4cba5d7
EK
39
40A guest operating system running in the emulated computer accesses these
41devices, and runs as it were running on real hardware. For instance you can pass
42an iso image as a parameter to Qemu, and the OS running in the emulated computer
189d3661 43will see a real CDROM inserted in a CD drive.
c4cba5d7 44
189d3661 45Qemu can emulates a great variety of hardware from ARM to Sparc, but {pve} is
c4cba5d7
EK
46only concerned with 32 and 64 bits PC clone emulation, since it represents the
47overwhelming majority of server hardware. The emulation of PC clones is also one
48of the fastest due to the availability of processor extensions which greatly
49speed up Qemu when the emulated architecture is the same as the host
9c63b5d9
EK
50architecture.
51
52NOTE: You may sometimes encounter the term _KVM_ (Kernel-based Virtual Machine).
53It means that Qemu is running with the support of the virtualization processor
54extensions, via the Linux kvm module. In the context of {pve} _Qemu_ and
55_KVM_ can be use interchangeably as Qemu in {pve} will always try to load the kvm
56module.
57
c4cba5d7
EK
58Qemu inside {pve} runs as a root process, since this is required to access block
59and PCI devices.
60
5eba0743 61
c4cba5d7
EK
62Emulated devices and paravirtualized devices
63--------------------------------------------
64
189d3661
DC
65The PC hardware emulated by Qemu includes a mainboard, network controllers,
66scsi, ide and sata controllers, serial ports (the complete list can be seen in
67the `kvm(1)` man page) all of them emulated in software. All these devices
68are the exact software equivalent of existing hardware devices, and if the OS
69running in the guest has the proper drivers it will use the devices as if it
c4cba5d7
EK
70were running on real hardware. This allows Qemu to runs _unmodified_ operating
71systems.
72
73This however has a performance cost, as running in software what was meant to
74run in hardware involves a lot of extra work for the host CPU. To mitigate this,
75Qemu can present to the guest operating system _paravirtualized devices_, where
76the guest OS recognizes it is running inside Qemu and cooperates with the
77hypervisor.
78
79Qemu relies on the virtio virtualization standard, and is thus able to presente
189d3661
DC
80paravirtualized virtio devices, which includes a paravirtualized generic disk
81controller, a paravirtualized network card, a paravirtualized serial port,
c4cba5d7
EK
82a paravirtualized SCSI controller, etc ...
83
189d3661
DC
84It is highly recommended to use the virtio devices whenever you can, as they
85provide a big performance improvement. Using the virtio generic disk controller
86versus an emulated IDE controller will double the sequential write throughput,
87as measured with `bonnie++(8)`. Using the virtio network interface can deliver
c4cba5d7 88up to three times the throughput of an emulated Intel E1000 network card, as
189d3661 89measured with `iperf(1)`. footnote:[See this benchmark on the KVM wiki
c4cba5d7
EK
90http://www.linux-kvm.org/page/Using_VirtIO_NIC]
91
5eba0743 92
80c0adcb 93[[qm_virtual_machines_settings]]
c4cba5d7
EK
94Virtual Machines settings
95-------------------------
80c0adcb 96
c4cba5d7
EK
97Generally speaking {pve} tries to choose sane defaults for virtual machines
98(VM). Make sure you understand the meaning of the settings you change, as it
99could incur a performance slowdown, or putting your data at risk.
100
5eba0743 101
80c0adcb 102[[qm_general_settings]]
c4cba5d7
EK
103General Settings
104~~~~~~~~~~~~~~~~
80c0adcb 105
c4cba5d7
EK
106General settings of a VM include
107
108* the *Node* : the physical server on which the VM will run
109* the *VM ID*: a unique number in this {pve} installation used to identify your VM
110* *Name*: a free form text string you can use to describe the VM
111* *Resource Pool*: a logical group of VMs
112
5eba0743 113
80c0adcb 114[[qm_os_settings]]
c4cba5d7
EK
115OS Settings
116~~~~~~~~~~~
80c0adcb 117
c4cba5d7
EK
118When creating a VM, setting the proper Operating System(OS) allows {pve} to
119optimize some low level parameters. For instance Windows OS expect the BIOS
120clock to use the local time, while Unix based OS expect the BIOS clock to have
121the UTC time.
122
5eba0743 123
80c0adcb 124[[qm_hard_disk]]
c4cba5d7
EK
125Hard Disk
126~~~~~~~~~
80c0adcb 127
2ec49380 128Qemu can emulate a number of storage controllers:
c4cba5d7
EK
129
130* the *IDE* controller, has a design which goes back to the 1984 PC/AT disk
131controller. Even if this controller has been superseded by more more designs,
132each and every OS you can think has support for it, making it a great choice
133if you want to run an OS released before 2003. You can connect up to 4 devices
134on this controller.
135
136* the *SATA* (Serial ATA) controller, dating from 2003, has a more modern
137design, allowing higher throughput and a greater number of devices to be
138connected. You can connect up to 6 devices on this controller.
139
b0b6802b
EK
140* the *SCSI* controller, designed in 1985, is commonly found on server grade
141hardware, and can connect up to 14 storage devices. {pve} emulates by default a
142LSI 53C895A controller. +
143A SCSI controller of type _Virtio_ is the recommended setting if you aim for
144performance and is automatically selected for newly created Linux VMs since
145{pve} 4.3. Linux distributions have support for this controller since 2012, and
c4cba5d7 146FreeBSD since 2014. For Windows OSes, you need to provide an extra iso
b0b6802b
EK
147containing the drivers during the installation.
148// https://pve.proxmox.com/wiki/Paravirtualized_Block_Drivers_for_Windows#During_windows_installation.
149
150* The *Virtio* controller, also called virtio-blk to distinguish from
151the Virtio SCSI controller, is an older type of paravirtualized controller
152which has been superseded in features by the Virtio SCSI Controller.
c4cba5d7
EK
153
154On each controller you attach a number of emulated hard disks, which are backed
155by a file or a block device residing in the configured storage. The choice of
156a storage type will determine the format of the hard disk image. Storages which
157present block devices (LVM, ZFS, Ceph) will require the *raw disk image format*,
158whereas files based storages (Ext4, NFS, GlusterFS) will let you to choose
159either the *raw disk image format* or the *QEMU image format*.
160
161 * the *QEMU image format* is a copy on write format which allows snapshots, and
162 thin provisioning of the disk image.
189d3661
DC
163 * the *raw disk image* is a bit-to-bit image of a hard disk, similar to what
164 you would get when executing the `dd` command on a block device in Linux. This
165 format do not support thin provisioning or snapshotting by itself, requiring
166 cooperation from the storage layer for these tasks. It is however 10% faster
167 than the *QEMU image format*. footnote:[See this benchmark for details
c4cba5d7 168 http://events.linuxfoundation.org/sites/events/files/slides/CloudOpen2013_Khoa_Huynh_v3.pdf]
189d3661 169 * the *VMware image format* only makes sense if you intend to import/export the
c4cba5d7
EK
170 disk image to other hypervisors.
171
172Setting the *Cache* mode of the hard drive will impact how the host system will
173notify the guest systems of block write completions. The *No cache* default
174means that the guest system will be notified that a write is complete when each
175block reaches the physical storage write queue, ignoring the host page cache.
176This provides a good balance between safety and speed.
177
178If you want the {pve} backup manager to skip a disk when doing a backup of a VM,
179you can set the *No backup* option on that disk.
180
181If your storage supports _thin provisioning_ (see the storage chapter in the
182{pve} guide), and your VM has a *SCSI* controller you can activate the *Discard*
183option on the hard disks connected to that controller. With *Discard* enabled,
184when the filesystem of a VM marks blocks as unused after removing files, the
185emulated SCSI controller will relay this information to the storage, which will
186then shrink the disk image accordingly.
187
af9c6de1 188.IO Thread
2b6e4b66
EK
189The option *IO Thread* can only be enabled when using a disk with the *VirtIO* controller,
190or with the *SCSI* controller, when the emulated controller type is *VirtIO SCSI*.
c564fc52
DC
191With this enabled, Qemu uses one thread per disk, instead of one thread for all,
192so it should increase performance when using multiple disks.
193Note that backups do not currently work with *IO Thread* enabled.
194
80c0adcb
DM
195
196[[qm_cpu]]
34e541c5
EK
197CPU
198~~~
80c0adcb 199
34e541c5
EK
200A *CPU socket* is a physical slot on a PC motherboard where you can plug a CPU.
201This CPU can then contain one or many *cores*, which are independent
202processing units. Whether you have a single CPU socket with 4 cores, or two CPU
203sockets with two cores is mostly irrelevant from a performance point of view.
204However some software is licensed depending on the number of sockets you have in
205your machine, in that case it makes sense to set the number of of sockets to
206what the license allows you, and increase the number of cores. +
207Increasing the number of virtual cpus (cores and sockets) will usually provide a
208performance improvement though that is heavily dependent on the use of the VM.
209Multithreaded applications will of course benefit from a large number of
210virtual cpus, as for each virtual cpu you add, Qemu will create a new thread of
211execution on the host system. If you're not sure about the workload of your VM,
212it is usually a safe bet to set the number of *Total cores* to 2.
213
214NOTE: It is perfectly safe to set the _overall_ number of total cores in all
215your VMs to be greater than the number of of cores you have on your server (ie.
2164 VMs with each 4 Total cores running in a 8 core machine is OK) In that case
217the host system will balance the Qemu execution threads between your server
218cores just like if you were running a standard multithreaded application.
219However {pve} will prevent you to allocate on a _single_ machine more vcpus than
220physically available, as this will only bring the performance down due to the
221cost of context switches.
222
223Qemu can emulate a number different of *CPU types* from 486 to the latest Xeon
224processors. Each new processor generation adds new features, like hardware
225assisted 3d rendering, random number generation, memory protection, etc ...
226Usually you should select for your VM a processor type which closely matches the
227CPU of the host system, as it means that the host CPU features (also called _CPU
228flags_ ) will be available in your VMs. If you want an exact match, you can set
229the CPU type to *host* in which case the VM will have exactly the same CPU flags
230as your host system. +
231This has a downside though. If you want to do a live migration of VMs between
232different hosts, your VM might end up on a new system with a different CPU type.
233If the CPU flags passed to the guest are missing, the qemu process will stop. To
234remedy this Qemu has also its own CPU type *kvm64*, that {pve} uses by defaults.
235kvm64 is a Pentium 4 look a like CPU type, which has a reduced CPU flags set,
236but is guaranteed to work everywhere. +
237 In short, if you care about live migration and moving VMs between nodes, leave
238the kvm64 default. If you don’t care about live migration, set the CPU type to
239host, as in theory this will give your guests maximum performance.
240
241You can also optionally emulate a *NUMA* architecture in your VMs. The basics of
242the NUMA architecture mean that instead of having a global memory pool available
243to all your cores, the memory is spread into local banks close to each socket.
244This can bring speed improvements as the memory bus is not a bottleneck
245anymore. If your system has a NUMA architecture footnote:[if the command
246`numactl --hardware | grep available` returns more than one node, then your host
247system has a NUMA architecture] we recommend to activate the option, as this
248will allow proper distribution of the VM resources on the host system. This
249option is also required in {pve} to allow hotplugging of cores and RAM to a VM.
250
251If the NUMA option is used, it is recommended to set the number of sockets to
252the number of sockets of the host system.
253
80c0adcb
DM
254
255[[qm_memory]]
34e541c5
EK
256Memory
257~~~~~~
80c0adcb 258
34e541c5
EK
259For each VM you have the option to set a fixed size memory or asking
260{pve} to dynamically allocate memory based on the current RAM usage of the
261host.
262
263When choosing a *fixed size memory* {pve} will simply allocate what you
264specify to your VM.
265
266// see autoballoon() in pvestatd.pm
267When choosing to *automatically allocate memory*, {pve} will make sure that the
268minimum amount you specified is always available to the VM, and if RAM usage on
269the host is below 80%, will dynamically add memory to the guest up to the
270maximum memory specified. +
271When the host is becoming short on RAM, the VM will then release some memory
272back to the host, swapping running processes if needed and starting the oom
273killer in last resort. The passing around of memory between host and guest is
274done via a special `balloon` kernel driver running inside the guest, which will
275grab or release memory pages from the host.
276footnote:[A good explanation of the inner workings of the balloon driver can be found here https://rwmj.wordpress.com/2010/07/17/virtio-balloon/]
277
c9f6e1a4
EK
278When multiple VMs use the autoallocate facility, it is possible to set a
279*Shares* coefficient which indicates the relative amount of the free host memory
280that each VM shoud take. Suppose for instance you have four VMs, three of them
281running a HTTP server and the last one is a database server. To cache more
282database blocks in the database server RAM, you would like to prioritize the
283database VM when spare RAM is available. For this you assign a Shares property
284of 3000 to the database VM, leaving the other VMs to the Shares default setting
285of 1000. The host server has 32GB of RAM, and is curring using 16GB, leaving 32
286* 80/100 - 16 = 9GB RAM to be allocated to the VMs. The database VM will get 9 *
2873000 / (3000 + 1000 + 1000 + 1000) = 4.5 GB extra RAM and each HTTP server will
288get 1/5 GB.
289
34e541c5
EK
290All Linux distributions released after 2010 have the balloon kernel driver
291included. For Windows OSes, the balloon driver needs to be added manually and can
292incur a slowdown of the guest, so we don't recommend using it on critical
293systems.
294// see https://forum.proxmox.com/threads/solved-hyper-threading-vs-no-hyper-threading-fixed-vs-variable-memory.20265/
295
296When allocating RAMs to your VMs, a good rule of thumb is always to leave 1GB
297of RAM available to the host.
298
80c0adcb
DM
299
300[[qm_network_device]]
1ff7835b
EK
301Network Device
302~~~~~~~~~~~~~~
80c0adcb 303
1ff7835b
EK
304Each VM can have many _Network interface controllers_ (NIC), of four different
305types:
306
307 * *Intel E1000* is the default, and emulates an Intel Gigabit network card.
308 * the *VirtIO* paravirtualized NIC should be used if you aim for maximum
309performance. Like all VirtIO devices, the guest OS should have the proper driver
310installed.
311 * the *Realtek 8139* emulates an older 100 MB/s network card, and should
312only be used when emulating older operating systems ( released before 2002 )
313 * the *vmxnet3* is another paravirtualized device, which should only be used
314when importing a VM from another hypervisor.
315
316{pve} will generate for each NIC a random *MAC address*, so that your VM is
317addressable on Ethernet networks.
318
af9c6de1
EK
319The NIC you added to the VM can follow one of two differents models:
320
321 * in the default *Bridged mode* each virtual NIC is backed on the host by a
322_tap device_, ( a software loopback device simulating an Ethernet NIC ). This
323tap device is added to a bridge, by default vmbr0 in {pve}. In this mode, VMs
324have direct access to the Ethernet LAN on which the host is located.
325 * in the alternative *NAT mode*, each virtual NIC will only communicate with
326the Qemu user networking stack, where a builting router and DHCP server can
327provide network access. This built-in DHCP will serve adresses in the private
32810.0.2.0/24 range. The NAT mode is much slower than the bridged mode, and
329should only be used for testing.
330
331You can also skip adding a network device when creating a VM by selecting *No
332network device*.
333
334.Multiqueue
1ff7835b 335If you are using the VirtIO driver, you can optionally activate the
af9c6de1 336*Multiqueue* option. This option allows the guest OS to process networking
1ff7835b
EK
337packets using multiple virtual CPUs, providing an increase in the total number
338of packets transfered.
339
340//http://blog.vmsplice.net/2011/09/qemu-internals-vhost-architecture.html
341When using the VirtIO driver with {pve}, each NIC network queue is passed to the
342host kernel, where the queue will be processed by a kernel thread spawn by the
343vhost driver. With this option activated, it is possible to pass _multiple_
344network queues to the host kernel for each NIC.
345
346//https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/7/html/Virtualization_Tuning_and_Optimization_Guide/sect-Virtualization_Tuning_Optimization_Guide-Networking-Techniques.html#sect-Virtualization_Tuning_Optimization_Guide-Networking-Multi-queue_virtio-net
af9c6de1 347When using Multiqueue, it is recommended to set it to a value equal
1ff7835b
EK
348to the number of Total Cores of your guest. You also need to set in
349the VM the number of multi-purpose channels on each VirtIO NIC with the ethtool
350command:
351
352`ethtool -L eth0 combined X`
353
354where X is the number of the number of vcpus of the VM.
355
af9c6de1 356You should note that setting the Multiqueue parameter to a value greater
1ff7835b
EK
357than one will increase the CPU load on the host and guest systems as the
358traffic increases. We recommend to set this option only when the VM has to
359process a great number of incoming connections, such as when the VM is running
360as a router, reverse proxy or a busy HTTP server doing long polling.
361
80c0adcb 362
685cc8e0
DC
363USB Passthrough
364~~~~~~~~~~~~~~~
80c0adcb 365
685cc8e0
DC
366There are two different types of USB passthrough devices:
367
368* Host USB passtrough
369* SPICE USB passthrough
370
371Host USB passthrough works by giving a VM a USB device of the host.
372This can either be done via the vendor- and product-id, or
373via the host bus and port.
374
375The vendor/product-id looks like this: *0123:abcd*,
376where *0123* is the id of the vendor, and *abcd* is the id
377of the product, meaning two pieces of the same usb device
378have the same id.
379
380The bus/port looks like this: *1-2.3.4*, where *1* is the bus
381and *2.3.4* is the port path. This represents the physical
382ports of your host (depending of the internal order of the
383usb controllers).
384
385If a device is present in a VM configuration when the VM starts up,
386but the device is not present in the host, the VM can boot without problems.
387As soon as the device/port ist available in the host, it gets passed through.
388
389WARNING: Using this kind of USB passthrough, means that you cannot move
390a VM online to another host, since the hardware is only available
391on the host the VM is currently residing.
392
393The second type of passthrough is SPICE USB passthrough. This is useful
394if you use a SPICE client which supports it. If you add a SPICE USB port
395to your VM, you can passthrough a USB device from where your SPICE client is,
396directly to the VM (for example an input device or hardware dongle).
397
80c0adcb
DM
398
399[[qm_bios_and_uefi]]
076d60ae
DC
400BIOS and UEFI
401~~~~~~~~~~~~~
402
403In order to properly emulate a computer, QEMU needs to use a firmware.
404By default QEMU uses *SeaBIOS* for this, which is an open-source, x86 BIOS
405implementation. SeaBIOS is a good choice for most standard setups.
406
407There are, however, some scenarios in which a BIOS is not a good firmware
408to boot from, e.g. if you want to do VGA passthrough. footnote:[Alex Williamson has a very good blog entry about this.
409http://vfio.blogspot.co.at/2014/08/primary-graphics-assignment-without-vga.html]
410In such cases, you should rather use *OVMF*, which is an open-source UEFI implemenation. footnote:[See the OVMF Project http://www.tianocore.org/ovmf/]
411
412If you want to use OVMF, there are several things to consider:
413
414In order to save things like the *boot order*, there needs to be an EFI Disk.
415This disk will be included in backups and snapshots, and there can only be one.
416
417You can create such a disk with the following command:
418
419 qm set <vmid> -efidisk0 <storage>:1,format=<format>
420
421Where *<storage>* is the storage where you want to have the disk, and
422*<format>* is a format which the storage supports. Alternatively, you can
423create such a disk through the web interface with 'Add' -> 'EFI Disk' in the
424hardware section of a VM.
425
426When using OVMF with a virtual display (without VGA passthrough),
427you need to set the client resolution in the OVMF menu(which you can reach
428with a press of the ESC button during boot), or you have to choose
429SPICE as the display type.
430
288e3f46
EK
431[[qm_startup_and_shutdown]]
432Automatic Start and Shutdown of Virtual Machines
433~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
434
435After creating your VMs, you probably want them to start automatically
436when the host system boots. For this you need to select the option 'Start at
437boot' from the 'Options' Tab of your VM in the web interface, or set it with
438the following command:
439
440 qm set <vmid> -onboot 1
441
442In some case you want to be able to fine tune the boot order of your VMs, for
443instance if one of your VM is providing firewalling or DHCP to other guest
444systems.
445For this you can use the following parameters:
446
447* *Start/Shutdown order*: Defines the start order priority. E.g. set it to 1 if
448you want the VM to be the first to be started. (We use the reverse startup
449order for shutdown, so a machine with a start order of 1 would be the last to
450be shut down)
451* *Startup delay*: Defines the interval between this VM start and subsequent
452VMs starts . E.g. set it to 240 if you want to wait 240 seconds before starting
453other VMs.
454* *Shutdown timeout*: Defines the duration in seconds {pve} should wait
455for the VM to be offline after issuing a shutdown command.
456By default this value is set to 60, which means that {pve} will issue a
457shutdown request, wait 60s for the machine to be offline, and if after 60s
458the machine is still online will notify that the shutdown action failed.
459
460Please note that machines without a Start/Shutdown order parameter will always
461start after those where the parameter is set, and this parameter only
462makes sense between the machines running locally on a host, and not
463cluster-wide.
076d60ae 464
8c1189b6 465Managing Virtual Machines with `qm`
dd042288 466------------------------------------
f69cfd23 467
dd042288 468qm is the tool to manage Qemu/Kvm virtual machines on {pve}. You can
f69cfd23
DM
469create and destroy virtual machines, and control execution
470(start/stop/suspend/resume). Besides that, you can use qm to set
471parameters in the associated config file. It is also possible to
472create and delete virtual disks.
473
dd042288
EK
474CLI Usage Examples
475~~~~~~~~~~~~~~~~~~
476
477Create a new VM with 4 GB IDE disk.
478
479 qm create 300 -ide0 4 -net0 e1000 -cdrom proxmox-mailgateway_2.1.iso
480
481Start the new VM
482
483 qm start 300
484
485Send a shutdown request, then wait until the VM is stopped.
486
487 qm shutdown 300 && qm wait 300
488
489Same as above, but only wait for 40 seconds.
490
491 qm shutdown 300 && qm wait 300 -timeout 40
492
f69cfd23
DM
493Configuration
494-------------
495
496All configuration files consists of lines in the form
497
498 PARAMETER: value
499
871e1fd6 500Configuration files are stored inside the Proxmox cluster file
8c1189b6 501system, and can be accessed at `/etc/pve/qemu-server/<VMID>.conf`.
f69cfd23 502
80c0adcb 503[[qm_options]]
a7f36905
DM
504Options
505~~~~~~~
506
507include::qm.conf.5-opts.adoc[]
508
f69cfd23
DM
509
510Locks
511-----
512
8c1189b6 513Online migrations and backups (`vzdump`) set a lock to prevent incompatible
871e1fd6
FG
514concurrent actions on the affected VMs. Sometimes you need to remove such a
515lock manually (e.g., after a power failure).
f69cfd23
DM
516
517 qm unlock <vmid>
518
f69cfd23
DM
519
520ifdef::manvolnum[]
521include::pve-copyright.adoc[]
522endif::manvolnum[]