首頁 » kvm

kvm 應用

由 datahunter 在 四, 07/06/2012 - 09:01 發表

最後更新: 2022-01-05

目錄

 

介紹

 

效能比較:

vbox vs kvm vs xen:

kvm > xen > vbox         # 使用 paravirtualization device 時

ie.

kvm: virtio

 

安裝

 

aptitude install qemu-kvm libvirt-bin           // 基本上 kvm 與 qemu 是密不可分的 !!

 

kernel module:

  • loadable kernel module: kvm.ko
  • processor specific module: kvm-intel.ko or kvm-amd.ko.

測試本機是否可用 kvm:

kvm-ok  <-- 主要是 check /dev/kvm 是否存在

INFO: Your CPU does not support KVM extensions
KVM acceleration can NOT be used

此機可以使用 kvm 技術

INFO: /dev/kvm exists
KVM acceleration can be used

注意 permission:

crw-rw---- 1 root kvm 10, 232 Jun  9 07:14 /dev/kvm

P.S.

系統能否用 KVM 技術, 取決於 CPU 上有沒有 HVM 功能(for Hardware Virtual Machine).

HVM 是指 CPU 支持虛擬技術的指令集, Intel 的指令集名叫 vt, AMD 的則是 svm

查看:

lsmod | grep kvm

kvm_intel             137721  0              <-- 每多一個 VM 時, 它會加一
kvm                   415549  1 kvm_intel

工具:

  • kvm                 <-- 它是 link 去 qemu-system-x86_64 package 的 (另有 qemu-i386, 它是 emulator)
  • kvm_stat         <-- qemu-kvm
  • kvm-img          <-- qemu-img package
  • kvm-nbd          <-- qemu-nbd package
  • kvm-ok            <-- cpu-checker package

 

KVM 支援的 VM image format

 

  • raw                <-- default format
  • qcow2 ( Encryption: AES-128bit (16 pw), Compression: zlib )
    backing_file, backing_fmt, encryption, cluster_size(512bytes -> 2Mbytes), preallocation(off|metadata)
  • qed
  • vdi (1.1)
  • vmdk (3,4)
  • cloop (Compressed Loop image)
  • cow                <-- 無用了
  • qcow
  • vpc

 

 

qemu-kvm 應用:

 

qemu-system-i386 與 qemu-i386 是有所不同的

qemu-system-i386 指的是 kvm

qemu-i386 則是 qemu 模擬器

 

qemu-kvm 使用:

 

建立 image:

qemu-img create -f qcow2 VM1.qcow2 5G

啟動:

qemu-system-i386  VM0.raw        <-- 最基本, 它會執行 VM0.raw

qemu-system-i386 -m 512 -vga std -vnc :0 -hda VM1.qcow -boot c

qemu-system-i386 -m 512 -vga std -vnc :0 -hda VM1.qcow -cdrom OS.iso -boot c -daemonize

 

-m megs                          <---- Default: 384 MB

-smp <n>                        <---- n 個 CPU Core

-hda/-hdb/-hdc/-hdd file   <-- 以IDE 槽掛 IMG

-boot:

  • d:  cdrom
  • c:  hard disk
  • n:  network

-vga

  • std
  • xenfb

-nographic: disable graphical output and redirect serial I/Os to console

Other Opts

  • -k language        # keyboard layout
  • -balloon none
  • -daemonize        # daemonize QEMU after initializing
  • -serial dev          #r edirect the serial port to char device 'dev'
  • -snapshot           # write to temporary files instead of disk image files

 

Linux/Multiboot boot specific

 

  • -kernel bzImage             use 'bzImage' as kernel image
  • -initrd file                        use 'file' as initial ram disk
  • -append cmdline             use 'cmdline' as kernel command line

 

VNC

 

入門:

-vnc :0    <-- 亦即是 tcp port 5900

進階:

-vnc 0.0.0.0:1,password   -k en-us

-k set the keyboard layout

 

Console:

 

Hotkey:

vnc 後, 可用以下 hotkey 進入不同的 console

[Ctrl] + [Alt] + [1]    console 1: VM 的 Display
[Ctrl] + [Alt] + [2]    console 2: QEMU monitor.
[Ctrl] + [Alt] + [3]    console 3: Serial output.
[Ctrl] + [Alt] + [4]    console 4: Parallel output.

 

QEMU monitor.

  • info history      <-- 鍵入過什麼
  • info kvm          <-- 是否在用 kvm 加速
  • info block         <-- 有什麼 block device
  • info blockstats
  • info cpus
  • info network
  • info vnc
  • info pci

 

換 CD:

(qemu) change ide1-cd0 /path/to/my.iso

(qemu) change ide1-cd0 /dev/sr0 host_device

eject ide1-cd0

 

USB:

usb_add disk:/tmp/disk.usb

(qemu)info usb

(qemu)usb_del 0.2

 

VNC Password:

set-password vnc YOU_PW

 

stop / cont / system_reset / system_powerdown

Quit

    quit or q

 

snapshots

  • qemu-snapshot Platte.img
  • commit hda
  • commit all
  • savevm name
  • info snapshots
  • loadvm name
  • commit device or commit all

 

詳見:

http://www.linux-kvm.org/page/Main_Page

 

Balloon driver & KSM

 

KSM (Kernel Sharedpage Merging) 是一種 memory overcommit 技術來,

原理是 de-duplication page, 一般它會與 ballooning 一起出現

KSM:

If any memory regions or pages are identical,

KSM reduces multiple identical memory pages to a single page. (page is then marked CoW)

Ballooning:

不同於 ksm 技術, 它是指在 VM 內安置特別的 Driver,

Ballooning 令到 VM 可以在執行期間改變 RAM 的容量

Remark:

Kernel >= 2.6.32 開始有

所有共享的 page 都是 write-protected page

查看主機是否支援 KSM:

grep KSM /boot/config-`uname -r`

CONFIG_KSM=y

/sys/kernel/mm/ksm# ls

run                                            <-- "1": 代表 ksmd 已被啟動; "0": stop ksmd from running but keep merged pages; "2": stop & unmerge pages
full_scans                                  <--  full scans 進行了多少次
pages_sharing                           <--  how many more sites are sharing the (how much saved)
pages_shared                            <-- Total pages shared. (how many shared pages are being used)
pages_unshared                        <-- how many pages unique but repeatedly checked for merging
pages_to_scan                         <--  每 N % 的 mergeable areas 被 scan 了完後 KSM proccess 才 sleep (sleep_millisecs) [Default: 100]
sleep_millisecs                           <-- milliseconds
pages_volatile                           <-- 有幾多 page 時常改變

 * A high ratio of pages_sharing to pages_shared indicates good sharing

 * A high ratio of pages_unshared to pages_sharing indicates wasted effort.

此外, 在系統上會見到有 [ksmd] 在執行工作

ballooning

(qemu) info balloon
(qemu) balloon 400   <-- vm 現在只可用 400 MB 了

Windows Virtio Drivers: (Redhat 出的)

  • ballooning memory <--- PCI standard RAM Controller
  • virtioserial
  • network
  • viostor ( block drivers )

下載:

http://alt.fedoraproject.org/pub/alt/virtio-win/latest/images/bin/

安裝 blnsvr (BalloonService):

  • blnsvr -i
  • blnsvr -u

 

Networking

 

-nic

A shortcut for configuring both the on-board guest NIC hardware and the host network backend in one go.

The host backend options are the same as with the corresponding -netdev options below.

Syntax

-nic                  # Default NIC with "user" host network backend

-nic none

-nic [tap|bridge|user|l2tpv3|vde|netmap|vhost-user|socket][,...][,mac=macaddr][,model=mn]

user-mode: (Default)

* requires no administrator privilege to run

-nic user,ipv6=off

Default 會以 DHCP 派 IP 比 Guest

IP 在 10.0.2.0/24 內                                                 # net=addr[/mask]

而 hypervisor 端的 IP 是 x.x.x.2 (ip link 是沒有它的)     # host=addr

ipv6=off  => Disable IPv6

port forward (hostfwd):

hostfwd=[tcp|udp]:[hostaddr]:hostport-[guestaddr]:guestport

hostaddr=0.0.0.0   # default

ie.

-nic user,ipv6=off,hostfwd=tcp:127.0.0.1:22001-:22

基本設定: model & mac

# model=e1000     (Default NIC)  # 查看支援的 model:  qemu-system-x86_64 -nic model=help

# mac=xx:xx:xx:yy:yy:yy          # 非必要

-nic mac=00:00:00:88:88:01

雙 NIC

-nic mac=52:54:00:88:88:01 -nic mac=52:54:00:88:88:02

Generate a random mac address

printf '52:54:00:88:%02X:%02X\n' $((RANDOM%256)) $((RANDOM%256))

tap bridged:

-nic tap,ifname=ros,mac=52:54:00:88:88:02

Script

,script=file,downscript=dfile

If name is not provided, the default network configure script is

  • /etc/qemu-ifup          # to add the tap device to the same bridge as we have default routing to.
  • /etc/qemu-ifdown

Use script=no or downscript=no to disable script execution.

My qemu-ifup script:

touch /etc/my-qemu-ifup

chmod 750 /etc/my-qemu-ifup

#!/bin/sh

SW=ros_br0

if [ -n "$1" ];then
  /sbin/ip link set $1 up
  sleep 1
  /sbin/ip link set $1 master $SW
  exit 0
else
  echo "Error: no interface specified"
  exit 1
fi

-nic tap,ifname=ros,mac=52:54:00:88:88:02,script=/etc/my-qemu-ifup

routing

Host:

sysctl -w net.ipv4.ip_forward=1                 # allow forwarding of IPv4

route add -host <ip-of-client> dev <tap-device> # add route to the client

VM:

route add -host <ip-of-host> dev <network-interface>

route add default gw <ip-of-host>

virtio:

... -nic model=virtio,mac= ...

 

Storage

 

The default storage driver: IDE

查看支援什麼 device

qemu-system-i386 -device ?

Controller/Bridge/Hub devices:
.............
Misc devices:
.............
name "virtio-balloon-device", bus virtio-bus
name "virtio-balloon-pci", bus PCI, alias "virtio-balloon"
name "virtio-rng-device", bus virtio-bus
name "virtio-rng-pci", bus PCI
.............
Uncategorized devices:
.............

cache 的方式:

  • writethrough (default) safe
  • cache=none

virtio:

qemu -drive file=/dev/mapper/ImagesVolumeGroup-Guest1,cache=none,if=virtio

 

CPU

 

限制 CPU 可用的指令集:

qemu -cpu qemu64,+ssse3,+sse4.1,+sse4.2,+x2apic

查看 CPU 類型及支援什麼指令集

qemu-system-x86_64 -cpu ?

Guest 可以用 host 上除 vms 外所有指令集:

qemu -cpu host,-vmx

* 好易出事, 如非必要, 勿用 !!

 

Default Config

 

-no-user-config

The "-no-user-config" option makes QEMU not load any of the user-provided config files on sysconfdir

-nodefaults

Don't create default devices. Normally, QEMU sets the default devices like serial port, parallel port, virtual console, monitor device, VGA adapter, floppy and CD-ROM drive and others.

 

Other Setting

 

-no-shutdown

Don't exit QEMU on guest shutdown, but instead only stop the emulation.  

This allows for instance switching to monitor to commit changes to the disk image.

-realtime mlock=on|off

Run qemu with realtime features.

mlocking qemu and guest memory can be enabled via mlock=on
(enabled by default)

 

OS 是否 VM

 

Windows

msinfo32

"System Manufacturer" and "System Model items"

 

VNC & Console

 

VNC

-vnc 192.168.88.150:9 -device cirrus-vga

Serial Console

-serial OPT

telnet:host:port[,server][,nowait][,nodelay]

The telnet protocol is used instead of raw tcp sockets.

If you use the server option QEMU will wait for a client socket application
to connect to the port before continuing, unless the "nowait" option was specified.

ie.

-serial telnet:localhost:23001,server,nowait

 

Time

 

-no-hpet

The High Precision Event Timer

-rtc

Specify base as "utc" or "localtime" to let the RTC start

-rtc base=utc

 

Daemonize

 

-daemonize       # Daemonize the QEMU process after initialization.

-pidfile file          # Store the QEMU process PID in file.

 

用 cmd start routeros

 

start.sh

#!/bin/bash

LIVECD=/mnt/raid/_iso/systemrescuecd/systemrescuecd-x86-4.9.2.iso
HDD=/kvm/vm/ros/ros.qcow2
SN=00000000000000000001
Model="VMware Virtual IDE Hard Drive"
PID=/kvm/vm/ros/ros.pid

/usr/bin/qemu-system-x86_64 -name guest=ROS \
 -machine pc-i440fx-3.1,accel=kvm,usb=off,vmport=off,dump-guest-core=off \
 -cpu SandyBridge -smp 1,sockets=1,cores=1,threads=1 \
 -m 512 -realtime mlock=off \
 -uuid 036a8e6e-adba-4887-8179-ce782dd2f2e4 \
 -boot order=c \
 -no-user-config -nodefaults \
 -rtc base=utc -no-hpet \
 -drive file=$HDD,format=qcow2,if=none,id=drive-ide0-0-0 \
 -device ide-hd,bus=ide.0,unit=0,drive=drive-ide0-0-0,id=ide0-0-0,serial=$SN,model="$Model" \
 -drive file=$LIVECD,format=raw,if=none,id=drive-ide0-1-0,readonly=on \
 -device ide-cd,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0 \
 -nic tap,ifname=ros-wan,model=virtio,mac=52:54:00:88:88:01,vhost=on \
 -nic tap,ifname=ros-lan,model=virtio,mac=52:54:00:88:88:02,script=/etc/my-qemu-ifup,vhost=on \
 -vnc 192.168.88.150:9 -device cirrus-vga \
 -serial telnet:localhost:23001,server,nowait \
 -daemonize -pidfile $PID

 * "-cpu host" 設定時 Guest 會 start 唔到

 

»

Creative Commons license icon Creative Commons license icon