[📖Getting started] [❓FAQs & Troubleshooting]
Lima launches Linux virtual machines with automatic file sharing, port forwarding, and containerd.
Lima can be considered as a some sort of unofficial "macOS subsystem for Linux", or "containerd for Mac".
Lima is expected to be used on macOS hosts, but can be used on Linux hosts as well.
✅ Automatic file sharing
✅ Automatic port forwarding
✅ Built-in support for containerd
✅ Intel on Intel
✅ ARM on Intel
✅ ARM on ARM
✅ Intel on ARM
✅ Various guest Linux distributions: Ubuntu, Debian, Fedora, Alpine, Arch Linux, openSUSE...
Related project: sshocker (ssh with file sharing and port forwarding)
This project is unrelated to The Lima driver project (driver for ARM Mali GPUs).
The goal of Lima is to promote containerd including nerdctl (contaiNERD ctl) to Mac users, but Lima can be used for non-container applications as well.
$ uname -a
Darwin macbook.local 20.4.0 Darwin Kernel Version 20.4.0: Thu Apr 22 21:46:47 PDT 2021; root:xnu-7195.101.2~1/RELEASE_X86_64 x86_64
$ lima uname -a
Linux lima-default 5.11.0-16-generic #17-Ubuntu SMP Wed Apr 14 20:12:43 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux
$ LIMA_INSTANCE=arm lima uname -a
Linux lima-arm 5.11.0-16-generic #17-Ubuntu SMP Wed Apr 14 20:10:16 UTC 2021 aarch64 aarch64 aarch64 GNU/Linux
$ echo "files under /Users on macOS filesystem are readable from Linux" > some-file
$ lima cat some-file
files under /Users on macOS filesystem are readable from Linux
$ lima sh -c 'echo "/tmp/lima is writable from both macOS and Linux" > /tmp/lima/another-file'
$ cat /tmp/lima/another-file
/tmp/lima is writable from both macOS and Linux"
$ lima nerdctl run -d --name nginx -p 127.0.0.1:8080:80 nginx:alpine
http://127.0.0.1:8080 is accessible from both macOS and Linux.
NOTE Privileged ports (1-1023) cannot be forwarded
For the usage of containerd and nerdctl (contaiNERD ctl), visit https://github.com/containerd/containerd and https://github.com/containerd/nerdctl.
Homebrew package is available for Intel Mac.
$ brew install lima
Manual installation steps (required for ARM Mac)
-
Install recent version of QEMU. v6.0.0 or later is recommended.
-
On ARM hosts, QEMU has to be patched for enabling
--accel=hvf
support. See https://gist.github.com/nrjdalal/e70249bb5d2e9d844cc203fd11f74c55 -
Download the binary archive of Lima from https://github.com/lima-vm/lima/releases , and extract it under
/usr/local
(or somewhere else). -
To install Lima from the source, run
make && make install
.
NOTE Lima is not regularly tested on ARM Mac.
[macOS]$ limactl start
...
INFO[0029] READY. Run `lima` to open the shell.
[macOS]$ lima uname
Linux
Detailed usage:
-
Run
limactl start <INSTANCE>
to start the Linux instance. The default instance name is "default". Lima automatically opens an editor (vi
) for reviewing and modifying the configuration. Wait until "READY" to be printed on the host terminal. -
Run
limactl shell <INSTANCE> <COMMAND>
to launch<COMMAND>
on Linux. For the "default" instance, this command can be shortened aslima <COMMAND>
. Thelima
command also accepts the instance name as the environment variable$LIMA_INSTANCE
. -
Run
limactl copy <SOURCE> ... <TARGET>
to copy files between instances, or between instances and the host. Use<INSTANCE>:<FILENAME>
to specify a source or target inside an instance. -
Run
limactl list [--json]
to show the instances. -
Run
limactl stop [--force] <INSTANCE>
to stop the instance. -
Run
limactl delete [--force] <INSTANCE>
to delete the instance. -
To enable bash completion, add
source <(limactl completion bash)
to~/.bash_profile
.
Lima may have bugs that result in loss of data.
Make sure to back up your data before running Lima.
Especially, the following data might be easily lost:
- Data in the shared writable directories (
/tmp/lima
by default), probably after hibernation of the host machine (e.g., after closing and reopening the laptop lid) - Data in the VM image, mostly when upgrading the version of lima
See ./pkg/limayaml/default.yaml
.
The current default spec:
- OS: Ubuntu 21.04 (Hirsute Hippo)
- CPU (x86_64): Haswell v4, 4 cores
- CPU (aarch64): Cortex A72, 4 cores
- Memory: 4 GiB
- Disk: 100 GiB
- Mounts:
~
(read-only),/tmp/lima
(writable) - SSH: 127.0.0.1:60022
- Hypervisor: QEMU with HVF accelerator
- Filesystem sharing: reverse sshfs (planned to be replaced with 9p soon)
- Port forwarding:
ssh -L
, automated by watching/proc/net/tcp
in the guest
-
Please certify your Developer Certificate of Origin (DCO), by signing off your commit with
git commit -s
and with your real name. -
Please squash commits.
🙏
- Test on ARM Mac
- Performance optimization
- More guest distros
- Windows hosts
- GUI with system tray icon (Qt or Electron, for portability)
- VirtFS to replace the current reverse sshfs (work has to be done on QEMU repo)
- vsock to replace SSH (work has to be done on QEMU repo)
Password is disabled and locked by default.
You have to use limactl shell bash
(or lima bash
) to open a shell.
Alternatively, you may also directly ssh into the guest: ssh -p 60022 -i ~/.lima/_config/user -o NoHostAuthenticationForLocalhost=yes 127.0.0.1
.
Yes, it should work, but not regularly tested on ARM.
Debian, Fedora, Alpine, Arch Linux, and openSUSE are also known to work.
See ./examples/
.
An image has to satisfy the following requirements:
- systemd or OpenRC
- cloud-init
- The following binaries to be preinstalled:
sudo
- The following binaries to be preinstalled, or installable via the package manager:
sshfs
newuidmap
andnewgidmap
apt-get
,dnf
,apk
,pacman
, orzypper
(if you want to contribute support for another package manager, rungit grep apt-get
to find out where to modify)
Yes, if you install it.
containerd can be stopped with systemctl --user disable --now containerd
.
Lima itself does not support connecting to a remote Linux machine, but sshocker, the predecessor or Lima, provides similar features for remote Linux machines.
e.g., run sshocker -v /Users/foo:/home/foo/mnt -p 8080:80 <USER>@<HOST>
to expose /Users/foo
to the remote machine as /home/foo/mnt
,
and forward localhost:8080
to the port 80 of the remote machine.
Lima is free software (Apache License 2.0), while Docker for Mac is not. Their EULA even prohibits disclosure of benchmarking result.
On the other hand, Moby, aka Docker for Linux, is free software, but Moby/Docker lacks several novel features of containerd, such as:
- On-demand image pulling (aka lazy-pulling, eStargz)
- Running an encrypted container
- Importing and exporting local OCI archives
If you have installed QEMU v6.0.0 or later on macOS 11 via homebrew, your QEMU binary should have been already automatically signed to enable HVF acceleration.
However, if you see HV_ERROR
, you might need to sign the binary manually.
cat >entitlements.xml <<EOF
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>com.apple.security.hypervisor</key>
<true/>
</dict>
</plist>
EOF
codesign -s - --entitlements entitlements.xml --force /usr/local/bin/qemu-system-x86_64
Note: Only on macOS versions before 10.15.7 you might need to add this entitlement in addition:
<key>com.apple.vm.hypervisor</key>
<true/>
- Make sure that HVF is enabled with
com.apple.security.hypervisor
entitlement. See "QEMU crashes withHV_ERROR
". - Emulating non-native machines (ARM-on-Intel, Intel-on-ARM) is slow by design.
- make sure qemu is codesigned, See "QEMU crashes with
HV_ERROR
". - if you are on macOS 10.15.7 or 11.0 or later make sure the entitlement
com.apple.vm.hypervisor
is not added. It only works on older macOS versions. You can clear the codesigning withcodesign --remove-signature /usr/local/bin/qemu-system-x86_64
and start over.
Privileged ports (1-1023) cannot be forwarded. e.g., you have to use 8080, not 80.
libslirp v4.6.0 used by QEMU is known to be broken.
If you have libslirp v4.6.0 in /usr/local/Cellar/libslirp
, you have to upgrade it to v4.6.1 or later (brew upgrade
).
The copy
command only works for instances that have been created by lima 0.5.0 or later. You can manually install the required identity on older instances with (replace INSTANCE
with actual instance name):
< ~/.lima/_config/user.pub limactl shell INSTANCE sh -c 'tee -a ~/.ssh/authorized_keys'
The default guest IP 192.168.5.15 is not accessible from the host and other guests.
To add another IP address that is accessible from the host and other virtual machines, enable vde_vmnet
.
See ./docs/network.md
.
- Inspect logs:
limactl --debug start
$HOME/.lima/<INSTANCE>/serial.log
/var/log/cloud-init-output.log
(inside the guest)/var/log/cloud-init.log
(inside the guest)
- Make sure that you aren't mixing up tabs and spaces in the YAML.