[en] This project provides three fool-proof scripts vm, ns, netns; designed to simplify the steps for QE and developers to configure and use VM/Container/netns, Thereby focusing more on the verification and testing of business functions.
[zh] 本项目提供了三个傻瓜化的脚本 vm, ns, netns ; 旨在简化测试和开发人员配置和使用 VM/Container/netns 的步骤， 从而更加聚焦业务功能的验证和测试。
vm:
vm create $distro -n $vmname [other options]
vm exec $vmname -- command line
ns:
ns create $ns [other options]
ns exec $ns -- command line
netns:
netns host,$vethX,$addr---$netns0,$vethX_peer,$addr $netns0,$vnic_ifname[,$addr][?updev=$if,mode=$mode]
netns exec $netns -- command line
[me@ws kiss-vm-ns]$ vm -h
Usage:
vm [create] <distro_or_family_name> [OPTIONs] ...
vm <$other_subcmd> [OPTIONs] ...
Options:
-h,--help #Display this help.
Options for sub-command create:
-I #create VM by import existing disk image, auto search url according distro name
-i <url/path> #create VM by import existing disk image, value can be url or local path
-L #create VM by using location, auto search url according distro name
-l <url/path> #create VM by using specified location url or local iso file path
-C <iso path> #create VM by using ISO image
--ks <file> #kickstart file, will auto generate according distro name if omitting
-n,--vmname <name>
#VM name, will auto generate according distro name if omitting
-f,--force #over write existing VM with same name
-p,--pkginstall <pkgs>
#pkgs in default system repo, install by yum/apt/zypper
-b,--brewinstall <args>
#pkgs in brew system or specified by url, install by internal brewinstall.sh
`-> just could be used in Intranet
-g,--genimage #generate VM image, after install shutdown VM and generate new qcow2.xz file
--rm #like --rm option of docker/podman, remove VM after quit from console
--nocloud,--nocloud-init
#don't create cloud-init iso for the image that is not cloud image
--osv <variant>
#OS_VARIANT, optional. virt-install will attempt to auto detect this value
# you can get [-osv variant] info by using:
$ osinfo-query os #RHEL-7 and later
$ virt-install --os-variant list #RHEL-6
--nointeract #exit from virsh console after install finish
--noauto #enter virsh console after installing start
--saveimage [path]
#save image in path if install with import mode
--downloadonly #download image only if there is qcow* image
--cpus <N> #number of virtual cpus, default 4
--msize <size> #memory size, default 2048
--dsize <size> #disk size, default 16
--net <$name[,$model]>
#attach tun dev(vnetN) and connect to net $name, optional $model: virtio,e1000,...
--net-br <$brname[,$model]>
#attach tun dev(vnetN) and connect to bridge $brname, optional $model: virtio,e1000,...
--net-macvtap,--netmacvtap [$sourceNIC[,$model]]
#attach macvtap interface over $sourceNIC, optional $model: virtio,e1000,...
--macvtapmode <vepa|bridge>
#macvtap mode
--xdisk <size[,fstype]>
#add an extra disk, could be specified multi-times. size unit is G
#`e.g: --xdisk 10 --xdisk 20,xfs
--disk <img[,bus=]>
#add exist disk file, could be specified multi-times.
--bus <$boot_disk_bus>
--sharedir <shpath[:target]>
#share path between host and guest
--nvdimm <nvdimm list>
#one or more nvdimm specification, format: 511+1 (targetSize+labelSize)
#`e.g: --nvdimm="511+1 1023+1" -> two nvdimm device
#`e.g: --nvdimm="511 1023" -> two nvdimm device
# ^^^^^^^^ default labelSize is 1, if omitting
#Note: will exit if qemu on your system does not support nvdimm, check by:
# PATH=$PATH:/usr/libexec qemu-kvm -device help | grep nvdimm
--nvme <size=[,format=]>
#one or more nvme specification.
#`e.g: --nvme=size=10 --nvme=size=20,format=raw
#size units: GB, default format is qcow2
#Note: will exit if qemu on your system does not support nvme, check by:
# PATH=$PATH:/usr/libexec qemu-kvm -device help | grep nvme
--vtpm #enable virtual tpm
--kdump #enable kdump
--fips #enable fips
--postrepo <name:url>
#add dnf/yum <repo> after install, only for CentOS/RHEL/Fedora
#`e.g: --postrepo=beaker-tasks:http://beaker.engineering.fedora.com/rpms
--nosshkey #don't inject sshkey
--debug #debug mode
--vncput-after-install <msg>
#send string or key event ASAP after virt-intall
--xml #just generate xml
--machine <machine type>
#specify machine type #get supported type by: qemu-kvm -machine help
--virt-install-opts #Pass-through virt-install options
--qemu-opts #Pass-through qemu options
--qemu-env #Pass-through qemu env[s]
--enable-guest-hypv #enable guest hypervisor, same as --qemu-opts="-cpu host,+vmx" or --qemu-opts="-cpu host,+svm"
#ref: https://www.linux-kvm.org/page/Nested_Guests
--disable-guest-hypv #disable guest hypervisor
--pxe #PXE install
#`e.g: vm fedora-32 -n f32 -net-macvtap -pxe --noauto -f
--diskless #diskless install
#`e.g: vm fedora-32 -n f32-diskless --net pxenet --pxe --diskless -f
-v,--verbose #verbose mode
-q #quiet mode, intend suppress the outputs of command yum, curl
Options for sub-command reboot:
-w,--wait #wait util the 22 port(sshd) is available after reboot
Options for sub-command exec:
-v,--verbose #verbose mode
-x[arg] #expected return code of sub-command exec, if doesn't match output test fail msg
#`e.g: -x or -x0 or -x1,2,3 or -x1,10,100-200
Options for sub-command vncproc:
--get,--vncget #get vnc screen and convert to text by gocr
--put,--vncput <msg>
#send string or key event to vnc server, could be specified multi-times
#`e.g: --put root --put key:enter --put password --put key:enter
--putln,--vncputln <msg>
#alias of: --put msg --put key:enter
Examples for create vm from distro-db (Intranet):
vm [create] # will enter a TUI show you all available distros that could auto generate source url
vm [create] RHEL-7.7 # install RHEL-7.7 from cloud-image(by default)
vm [create] RHEL-6.10 -L # install RHEL-6.10 from Location(by -L option)
vm [create] RHEL-8.1.0 -f -p "vim wget git" # -f force install VM and ship pkgs: vim wget git
vm [create] RHEL-8.1.0 -brewinstall 23822847 # ship brew scratch build pkg (by task id)
vm [create] RHEL-8.1.0 -brewinstall kernel-4.18.0-147.8.el8 # ship brew build pkg (by build name)
vm [create] RHEL-8.1.0 -brewinstall "lstk -debug" # ship latest brew build release debug kernel
vm [create] RHEL-8.1.0 -brewinstall "upk -debug" # ship latest brew build upstream debug kernel
vm [create] RHEL-8.1.0 --nvdimm "511 1022+2" # add two nvdimm device
vm [create] RHEL-8.3.0 --nvme "size=32 size=16,format=raw" # add two nvme device
vm [create] rhel-8.2.0% # nightly 8.2 # fuzzy search distro: ignore-case
vm [create] rhel-8.2*-????????.? # rtt 8.2 # - and only support glob * ? syntax, and SQL %(same as *)
vm [create] rhel-8.2% -enable-guest-hypv -msize=$((8*1024)) -dsize=120 # enable hyper-v on guest
Examples for create vm from distro-db (Internet):
vm [create] # will enter a TUI show you all available distros that could auto generate source url
vm [create] CentOS-8-stream -b ftp://url/path/x.rpm
vm [create] CentOS-8 -p "jimtcl vim git make gcc"
vm [create] CentOS-7 -p "vim git wget make gcc"
vm [create] FreeBSD-13.0
vm [create] debian-10
vm [create] openSUSE-leap-15.2
vm [create] fedora-32
vm [create] CentOS-8 -enable-guest-hypv -msize=$((8*1024)) -dsize=120 # enable hyper-v on guest
Examples for create vm from local image or url:
vm [create] freebsd-130 -i ~/myimages/download/FreeBSD-13.0-RELEASE-amd64.qcow2.xz --nocloud-init
vm [create] debian-10 -i /mnt/vm-images/debian-10-openstack-amd64.qcow2
vm [create] openSUSE-15.2 -i ~/myimages/openSUSE-Leap-15.2-OpenStack.x86_64.qcow2
vm [create] debian-test -i https://cdimage.debian.org/cdimage/openstack/testing/debian-testing-openstack-amd64.qcow2
vm [create] centos-8 -l https://cloud.centos.org/centos/8/x86_64/images/
vm [create] rhel-8.4.0 -l ~/myimages/download/RHEL-8.4.0-any-x86_64-dvd1.iso -n rhel84-from-iso
Examples for other sub-commands:
vm prepare #check/install/configure libvirt and other dependent packages
vm enable-nested-vm #enable nested on host
vm list #list all VMs //you can use ls,li,lis* instead list
vm login [VM] #login VM via ssh //you can use l,lo,log* instead login
vm console [VM] #log VM via console //you can use co,con,cons* instead console
vm delete [VM list] #delete VMs //you can use d,de,del*,r,rm instead delete
vm ifaddr [VM] #show ip address //you can use i,if,if* instead ifaddr
vm vnc [VM] #show vnc host:port //you can use v,vn instead
vm vnc [-get|-put|-putln] [VM] #read screen text or send string thru vnc
vm xml [VM] #dump vm xml file //you can use x,xm instead xml
vm edit [VM] #edit vm xml file //you can use ed,ed* instead edit
vm exec [-v] [-x] "$VM" -- "cmd" #login VM and exec cmd //you can use e,ex,ex* instead exec
vm reboot [-w] [VM] #reboot VM //option /w indicate wait until reboot complete(port 22 is available)
vm stop [VM] #stop/shutdonw VM //nil
vm start [VM] #start VM //nil
vm cpfrom <VM> <file/dir_in_vm> <dst_dir/file_in_host>
vm cpto <VM> <files/dirs_in_host ...> <dst_dir_in_vm>
vm netls #list all virtual network
vm netcreat netname=nat-net brname=virbrM subnet=10 [forward=nat] #create network 'nat-net' with 'nat' and subnet: 192.168.10.0
vm netcreat netname=isolated-net brname=virbrN subnet=20 forward=no #create network 'isolated-net' with subnet: 192.168.20.0
vm netcreat netname=pxe brname=virpxebrN subnet=172.25.250.0 tftproot=/var/lib/tftpboot bootpfile=pxelinux/pxelinux.0
vm netinfo netname #show detail info of virtual network 'netname'
vm netstart netname #start virtual network 'netname'
vm netdel netname #delete virtual network 'netname'
[me@ws kiss-vm-ns]$ ns -h
Usage:
ns <-n nsname> [options] [exec -- cmdline | ps | del | install pkgs | {jj|jinja} pkgs ]
ns ls
Options:
-h, --help ; show this help info
-n {nsname} ; ns(name space) name
-p {pkgs} ; packages you want in ns(name space)
-d ; debug mode
-v ; verbose mode, output more info
--veth-ip {ip1,ip2} ; ip address pair for veth pair; ip1 for host side and ip2 for ns side
--macvlan-ip {ip1[,ip2...]} ; ip address[es] for ns macvlan if[s]; all for ns side
--bind {src[:dst]} ; see systemd-nspawn --bind
--robind {src[:dst]} ; see systemd-nspawn --bind-ro
--vol, --volatile {yes|no} ; see systemd-nspawn --volatile. default is no
--clone {ns} ; clone from ns
--noboot ; no boot
-x[arg] ; expected return code of sub-command exec, if doesn't match output test fail msg
; e.g: -x or -x0 or -x1,2,3 or -x1,10,100-200
Examples create ns by using mini fs tree + host /usr:
# same as example ns1, but use a it's own fs tree instead reuse host os tree
# so you can do anything in this ns, and don't worry about any impact on the host
ns jj nsmini bash # create rootfs template nsmini
ns -n ns0 --veth-ip 192.168.0.1,192.168.0.2 --noboot -robind=/usr --clone nsmini
ns -n ns1 --veth-ip 192.168.1.1,192.168.1.2 --macvlan-ip 192.168.254.11 -bind=/usr --clone nsmini
Examples create ns by using absolute own fs tree:
ns jj nsbase iproute iputils nfs-utils --clone nsmini # create rootfs template nsbase
ns -n ns2 --veth-ip 192.168.2.1,192.168.2.2 --macvlan-ip 192.168.254.12,192.168.253.12 --clone nsbase
ns -n ns3 --veth-ip 192.168.3.1,192.168.3.2 --macvlan-ip 192.168.254.13,192.168.253.13 --clone nsbase
Examples sub-command:
ns ls # list all ns
ns ps ns3 # show ps tree of ns3
ns del ns3 # delete/remove ns3 but keep rootdir
ns delete ns3 # delete/remove ns3 and it's rootdir
ns install ns2 cifs-utils # install cifs-utils in ns2
ns exec -v -x ns2 ip addr show # exec command in ns2
ns exec -v -x ns2 -- ls -l / # exec command in ns2
sudo systemctl start nfs-server
sudo exportfs -o ro,no_root_squash "*:/usr/share"
sudo addmacvlan macvlan-host
sudo addressup macvlan-host 192.168.254.254
sudo firewall-cmd --add-service=nfs --add-service=mountd --add-service=rpc-bind
ns exec -v -x ns2 -- mkdir -p /mnt/nfs # exec command in ns2
ns exec -v -x ns2 -- showmount -e 192.168.2.1 # exec command in ns2
ns exec -v -x ns2 -- mount 192.168.2.1:/ /mnt/nfs # exec command in ns2
ns exec -v -x ns2 -- showmount -e 192.168.254.254 # exec command in ns2
ns exec -v -x ns2 -- mount 192.168.254.254:/ /mnt/nfs # exec command in ns2
example1: https://github.com/tcler/linux-network-filesystems/blob/master/testcases/nfs/nfs-stress/nfs-stress.sh#L85
example2: https://github.com/tcler/linux-network-filesystems/blob/master/testcases/nfs/multipath/multipath-netns.sh#L23
[me@ws kiss-vm-ns]$ netns
Usage:
netns <$nsname,$vethX,$addr---$nsname,$vethX_peer,$addr | $nsname,$vnic_name[,$addr][?updev=$if,mode=$mode,iftype=$iftype]>
# ^^^ nsname 'host' means default network namespace, br-* means it's a bridge //[convention over configuration]
# ^^^ vnic_name 'iv-*' means ipvlan nic, 'mv-*' and others means macvlan nic //[convention over configuration]
# +--------+ +--------+
# | ns0 [veth0.X]------------[veth0.Y] host |
# +--------+ +--------+
# netns ns0,veth0.X,192.168.1.2---host,veth0.Y,192.168.1.1
# +--------+ +--------+
# | ns0 [veth1.X]------------[veth1.Y] ns1 |
# +--------+ +--------+
# netns ns0,veth1.X,192.168.2.2---ns1,veth1.Y,192.168.2.1
# +--------+ +------+ +--------+
# | ns0 [veth3.X]----[veth3.Y] br-0 [veth4.X]----[veth4.Y] ns1 |
# +--------+ +------+ +--------+
# netns ns0,veth3.X,192.168.3.2---br-0,veth3.Y br-0,veth4.X---ns1,veth4.Y,192.168.3.1
# +--------+ +--------+
# | [veth5.X]------------[veth5.Y] |
# | ns0 | | ns1 |
# | [mv-ns0] [mv-ns1] |
# +--------+ \ / +--------+
# \ /
# +------------------+
# | mv-ns0 | mv-ns1 |
# +------------------|
# | eth0 |
# +------------------+
# netns ns0,veth5.X,192.168.4.2---ns1,veth5.Y,192.168.4.1 ns0,mv-ns0,192.168.5.10 ns1,mv-ns1,192.168.5.11
netns exec $nsname -- cmdline
netns del $nsname
netns ls
netns veth ve0.a-host,ve0.b-ns0 #create veth pair
netns macvlan ifname #create macvlan if; [updev=updev] [mode={bridge|vepa|private|passthru}]
netns ipvlan ifname #create ipvlan if; [updev=updev] [mode={l2|l3}]
netns addrup $if $address #set address and up if
netns addif2netns $ns $if [$addr] #add new if to netns, [and setup address and up]
netns detach $ns $if #detach if from netns
Options:
-h, --help ; show this help info
-v ; verbose mode
-n <arg> ; netns name
-x[arg] ; expected return code of sub-command exec, if doesn't match output test fail msg
; e.g: -x or -x0 or -x1,2,3 or -x1,10,100-200
Examples: host connect ns0 with both veth and macvlan
netns host,ve0.a-host,192.168.0.1---ns0,ve0.b-ns0,192.168.0.2 host,mv-host0,192.168.100.1 ns0,mv-ns0,192.168.100.2
netns exec -v -x ns0 -- ping -c 2 192.168.0.1
netns exec -v -x ns0 -- ping -c 2 192.168.100.1
#curl -s -L https://raw.githubusercontent.com/tcler/linux-network-filesystems/master/tools/configure-nfs-server.sh | sudo bash
sudo systemctl start nfs-server
sudo exportfs -o ro,no_root_squash "*:/usr/share"
sudo firewall-cmd --add-service=nfs --add-service=mountd --add-service=rpc-bind
netns exec -v -x ns0 -- showmount -e 192.168.0.1
netns exec -v -x ns0 -- mkdir -p /mnt/ns0/nfs
netns exec -v -x ns0 -- mount 192.168.0.1:/ /mnt/ns0/nfs
netns exec -v -x ns0 -- mount -t nfs4
netns exec -v ns0 -- ls /mnt/ns0/nfs
netns exec -v -x ns0 -- umount /mnt/ns0/nfs
netns exec -v -x32 ns0 -- umount /mnt/ns0/nfs
netns exec -v -x ns0 -- umount /mnt/ns0/nfs #just for show what does -x option work, when test fail
netns exec -v ns0 -- rm -rf /mnt/ns0
netns del ns0
netns delif mv-host0
Examples: host connect ns0 with both veth and ipvlan
netns host,ve0.a-host,192.168.0.1---ns0,ve0.b-ns0,192.168.0.2 host,iv-host0,192.168.99.1 ns0,iv-ns0,192.168.99.2
netns exec -v -x ns0 -- ping -c 2 192.168.99.1
netns del ns0
netns delif iv-host0
Examples: host connect ns0 with veth and bridge br-0
netns host,veth0.X,192.168.66.1---br-0,veth0.Y br-0,veth1.Y---ns0,veth1.X,192.168.66.2
netns exec -v -x ns0 -- ping -c 2 192.168.66.1
netns del ns0 br-0