Warning: after this process you'll need to rebuild any lando projects that you want to use. If you have any information in previously built Lando projects you need to fetch you should do so before making the changes in the post but you can also: undo the changes to Lando's config.yml file to regain have access to your preexisting projects again.
Install colima, start a VM. More info https://ddev.readthedocs.io/en/latest/users/usage/faq/#features-requirements
brew install colima docker
colima start --cpu 4 --memory 6 --disk 100 --vm-type=qemu --mount-type=sshfs --dns=1.1.1.1
Once Colima is running check that your docker command uses it as a default backend
docker context ls | grep colima
colima * moby colima unix:///Users/duker/.colima/default/docker.sock
If you see the * next to colima you are ready to go forward, otherwise change docker to use colima as a backend, docker context will give you some help on that.
We need to configure Lando to use the port and socket from colima. The socket we can get from the previously ran command. To get the port we need to get a little more creative. We know that colima uses qemu for the VM it runs so we can list all processes listening on ports with
lsof -i | grep LISTEN | grep qemu
qemu-syst 44196 duker 25u IPv4 0x567251ae0f9e6083 0t0 TCP localhost:53590 (LISTEN)
The problem with this approach is that if you have more than one qemu VM running you wont be able to tell them appart, in that case you can do the right thing and get the port number from colima's ssh config with
❯ colima ssh-config | grep Port
Port 53590
Now that we know our socket (/Users/duker/.colima/default/docker.sock) and port number (53590) we can setup Lando to use the colima VM instead of Docker desktop for its containers, add the following section to your ~/.lando/config.yml:
engineConfig:
host: 127.0.0.1
port: 53590
socketPath: /Users/duker/.colima/default/docker.sock
After that step if all went well you can start using lando normally.
Some small benchmarks, each one was ran with no browser cache and using the Page Load Time Chrome extension on Brave.
The tests were ran on a Mac Studio 2022 with M1 Max and 64 GB of ram.
| URL | Page load time | Cached? | Backend |
|---|---|---|---|
| http://history-channel.lndo.site/ | 11.9s | No | Docker |
| http://history-channel.lndo.site/ | 1.84s | Yes | Docker |
| http://history-channel.lndo.site/ | 9.10s | No | Colima |
| http://history-channel.lndo.site/ | 1.60s | Yes | Colima |
| http://laemmle.lndo.site/ | 6.07s | No | Docker |
| http://laemmle.lndo.site/ | 1.64s | Yes | Docker |
| http://laemmle.lndo.site/ | 7.45s | No | Colima |
| http://laemmle.lndo.site/ | 2.20s | Yes | Colima |
Next steps try ddev and compare that too? https://ddev.readthedocs.io/en/stable/users/install/ddev-installation/
sudo apk add nodejs yarn docker-compose
git clone https://github.com/lando/cli.git lando-cli
cd lando-cli && yarn
sudo ln -s $(pwd)/bin/lando.js /usr/local/bin/lando.dev
cd && touch .profile
sudo apk add --no-cache libstdc++ coreutils curl bash
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.35.1/install.sh | bash
echo "export NVM_NODEJS_ORG_MIRROR=https://unofficial-builds.nodejs.org/download/release" >> .profile
echo "nvm_get_arch() { nvm_echo \"x64-musl\"; }" >> .profile
source .profile