madpah
commented
10 months ago I've had to revise my solution to run this command inside the colima VM (after sudo):
ip address add 192.168.106.2/24 broadcast + dev col0
Open liaden opened 2 years ago
madpah
commented
10 months ago I've had to revise my solution to run this command inside the colima VM (after sudo):
ip address add 192.168.106.2/24 broadcast + dev col0
norrs
commented
8 months ago So there seems to be multiple issues in the project regarding regarding creating a colima VM and successfully obtain a IPv4 address after --network-address is enabled during creation.
@norrs doing some research:
After reading source code for colima, and trying a million things mentions in all various issues.. I find hint from the comment here in this issue https://github.com/abiosoft/colima/issues/458#issuecomment-1741879608 which lead me on path to research if VM actaully got an IPv4 address from the dhcp server and if there was possible firewall issues. This lead me to the great issue over at
lima-vmat https://github.com/lima-vm/lima/issues/1259 and further into https://github.com/lima-vm/socket_vmnet/issues/18#issuecomment-1574149506 which seems to confirm my suspect.
col0 that is reachable from the host when using virtual machine type vz ( $ colima start --vm-type vz --network-address ).qemu ($ colima start --arch=x86_64 --network-address) 😢
You really want a firewall and not disable it, and
bootpdshould not really be blocked according to the description here, so this sounds like a Apple bug as mentioned in the socket_vmnet issue.
Executing the following commands below would add a firewall rule for bootpd and allow the VM to obtain an IPv4 address over DHCP.
$ sudo /usr/libexec/ApplicationFirewall/socketfilterfw --add /usr/libexec/bootpd
$ sudo /usr/libexec/ApplicationFirewall/socketfilterfw --unblock /usr/libexec/bootpdHowever, you might be blocked due to MDM (mobile device management) in your company if you are unlucky: https://github.com/lima-vm/socket_vmnet/issues/18#issuecomment-1717216573 and https://github.com/lima-vm/socket_vmnet/issues/18#issuecomment-1727195929 .
This issue seems to already be documented in lima-vm/socket_vmnet:README.
FYI @abiosoft - hope it helps clearing up errors others are experiencing.
For everyone else: This is also a good link for introduction on how lima under the hood does networking: https://lima-vm.io/docs/config/network/
And my background story and 😴 researching:
I had this working on Friday, I believe I did a required system upgrade due to policy on the computer during the weekend (and probably had colima running in the background) and ended up with Sonoma 14.4 on my newly freshly aquired Mac Pro M3. After reboot I noticed my VMs under colima did not receive an IPv4 addresses.
Silvanoc
commented
7 months ago It is sort of weird that I'm observing something completely different from what @norrs is reporting in this comment. I wonder if we're observing completely different issues or the same issue with different manifestations due to slight differences in our set-ups.
These are the observed differences.
Everything works and it obtains an IPv4 address inside the VM on interface col0 that is reachable from the host when using virtual machine type vz ( $ colima start --vm-type vz --network-address ).
In my case for VZ fails because the interface col0 is not even being created...
It sadly fails for qemu ($ colima start --arch=x86_64 --network-address)
In my case for QEMU it works perfectly.
My setup:
sudeeshjohn
commented
6 months ago It is sort of weird that I'm observing something completely different from what @norrs is reporting in this comment. I wonder if we're observing completely different issues or the same issue with different manifestations due to slight differences in our set-ups.
same here. IP address is not showing up for colima start --network-address --arch x86_64 --vm-type vz
greycr0w
commented
5 months ago M1 Sonoma 14.5 On my side, this was introduced when I stopped colima to test something out.
I was able to fix it with the following steps:
brew install --HEAD colimaSilvanoc
commented
5 months ago After having reported here that colima start --arch=x86_64 --network-address was working fine for me, now it has stopped working 🤦
Silvanoc
commented
5 months ago @abiosoft I have just tried brew install --head colima as you propose in a quite old comment. So I'm using now c3a31ed and I am still having difficulties to run colima start --arch=x86_64 --network-address on an M1 MacBook.
The reported error message is error calling fd_connect: fd_connect: dial unix /Users/[...]/.colima/test-net/daemon/vmnet.sock: connect: connection refused.
kaiserbrito
commented
2 months ago @abiosoft I have just tried
brew install --head colimaas you propose in a quite old comment. So I'm using now c3a31ed and I am still having difficulties to runcolima start --arch=x86_64 --network-addresson an M1 MacBook.The reported error message is
error calling fd_connect: fd_connect: dial unix /Users/[...]/.colima/test-net/daemon/vmnet.sock: connect: connection refused.
I was having the same issue, I followed the steps @abiosoft mentioned before:
brew uninstall colima qemu lima docker docker-compose
brew autoremove
brew cleanup
rm -rf ~/.docker ~/.colima
brew install colima qemu lima docker docker-compose
davidreghay
commented
2 months ago Seems like this is quite a persistent issue. Yesterday I was trying to follow @madpah's suggested workaround as given above here but on newer versions of lima running newer a newer release of Ubuntu the solution doesn't work since Ubuntu switched over to netplan.io.
I was able to update the col0 device address on the colima instance by adapting those instructions as follows (diverges at step 3):
(Wait for it to start) I confirm it's started with no network address by running:
colima status
INFO[0000] colima is running using macOS Virtualization.Framework
INFO[0000] arch: aarch64
INFO[0000] runtime: docker
INFO[0000] mountType: virtiofs
INFO[0000] address:
INFO[0000] socket: unix:///Users/username/.colima/default/docker.sockOn my Mac, there should now be a network bridge adapter created - mine is always bridge100 - grab it's IP address by running:
ifconfig bridge100 | grep "inet "
inet 192.168.106.1 netmask 0xffffff00 broadcast 192.168.106.255
colima ssh(you are now in SSH inside VM)
sudo su -Edit the file /etc/netplan/50-cloud-init.yaml so that it looks similar to the below, using the IP address returned in step 2 as the value for gateway (don't alter the eth0 settings or anything for col0 besides the addresses field):
# This file is generated from information provided by the datasource. Changes
# to it will not persist across an instance reboot. To disable cloud-init's
# network configuration capabilities, write a file
# /etc/cloud/cloud.cfg.d/99-disable-network-config.cfg with the following:
# network: {config: disabled}
network:
ethernets:
col0:
addresses:
- 192.168.64.1/24
match:
macaddress: 52:55:55:54:9b:81
set-name: col0
eth0:
dhcp4: true
match:
macaddress: 52:55:55:a5:45:37
nameservers:
addresses:
- 192.168.5.2
set-name: eth0
version: 2netplan applyThis worked for me but of course is a stopgap solution rather than longterm one.
@jamhed and @norrs your notes make sense but I'm not clear on how to apply them. Notably though, the manual fix as outlined above does not require having permission to change any firewall or networking rules on the host machine. Maybe the best thing would be supplying a configuration that allows starting it up with assigning col0 a static address instead of trying to use DHCP?
jubr
commented
1 month ago Lemme add my $0.02 as I am also a long-time sufferrer from --network-address-less-ness.
I finally decided to take the plunge, un-freeze my 0.5.6 & give it a go!
On M1 / Sonoma 14.7 / colima 0.7.5 / lima 0.23.2 / qemu 9.1.0 here. Applied socketfilterfw steps above.
I was watching tail -F /Users/jbraam/.colima/x86/daemon/daemon.log when I noticed:
time="2024-10-22T02:31:09+02:00" level=trace msg="cmd int [\"sudo\" \"/opt/colima/bin/socket_vmnet\" \"--vmnet-mode\" \"shared\" \"--socket-group\" \"staff\" \"--vmnet-gateway\" \"192.168.106.1\" \"--vmnet-dhcp-end\" \"192.168.106.254\" \"--pidfile\" \"/opt/colima/run/vmnet-x86.pid\" \"/Users/jbraam/.colima/x86/daemon/vmnet.sock\"]"
sudo: a terminal is required to read the password; either use the -S option to read from standard input or configure an askpass helper
sudo: a password is required
time="2024-10-22T02:31:09+02:00" level=error msg="error starting vmnet: error running vmnet: exit status 1"So it might look like the sudoers might be not working?
$ cat /etc/sudoers.d/colima
# starting vmnet daemon
%staff ALL=(root:wheel) NOPASSWD:NOSETENV: /opt/colima/bin/socket_vmnet --vmnet-mode shared --socket-group staff --vmnet-gateway 192.168.106.1 --vmnet-dhcp-end 192.168.106.254 *
# terminating vmnet daemon
%staff ALL=(root:wheel) NOPASSWD:NOSETENV: /usr/bin/pkill -F /opt/colima/run/*.pid
# validating vmnet daemon
%staff ALL=(root:wheel) NOPASSWD:NOSETENV: /usr/bin/pkill -0 -F /opt/colima/run/*.pidPerhaps this is because I am on a corp laptop.
Anyhoozers, I thought to try:
sudo /opt/homebrew/bin/colima daemon start x86 --vmnet --very-verboseand then:
colima start -p x86 --very-verbosewhich gave all the "regular" info, and finally:
$ colima ls
PROFILE STATUS ARCH CPUS MEMORY DISK RUNTIME ADDRESS
arm Stopped aarch64 3 6GiB 60GiB
x86 Running x86_64 2 3GiB 60GiB docker 192.168.106.2🎉 Oh happy day, and there was much rejoicing!
I hope this might take this by now epic saga one step further 😜
Description
Attempting to start
--network-addressfails. Additional starts without the flag fail with a different error message. Removing~/.colimaallows for a successful start, without the network address.Version
Colima Version: 0.4.6 Lima Version: 0.12.0, 0.13.0 Qemu Version: 7.1.0
Operating System
Reproduction Steps
colima start --network-addressExpected behaviour
Colima starts correctly.
Additional context
First start:
ha.stderr.log:
Second start:
which is obviously due to the colima.yaml being updated with the
--network-addressflag.