M1 Mac - multipass error when installing via script about dotrun HOT 11 CLOSED

Seems that while M1 Pro has 10 cores, only 8 are "high performance" cores available for virtualisation (my understanding).

In sysctl -a we can see:

hw.ncpu: 10
… 
hw.perflevel0.physicalcpu: 8
hw.perflevel0.physicalcpu_max: 8
hw.perflevel0.logicalcpu: 8
hw.perflevel0.logicalcpu_max: 8

On older Intel Mac:

hw.ncpu: 8
…
hw.perflevel0.physicalcpu: 4
hw.perflevel0.physicalcpu_max: 4
hw.perflevel0.logicalcpu: 8
hw.perflevel0.logicalcpu_max: 8

from dotrun.

Changing
https://github.com/canonical-web-and-design/dotrun/blob/a1943631447c5b17537150a219b025027263ea3c/scripts/install-with-multipass.sh#L56

to:

TOTAL_CPUS=$(sysctl -n hw.perflevel0.logicalcpu)

seems to work, but I don't know how backwards compatible it is with other Macs.

from dotrun.

@jkfran can you help triage this one pls

from dotrun.

Changing

https://github.com/canonical-web-and-design/dotrun/blob/a1943631447c5b17537150a219b025027263ea3c/scripts/install-with-multipass.sh#L56

to:

TOTAL_CPUS=$(sysctl -n hw.perflevel0.logicalcpu)

seems to work, but I don't know how backwards compatible it is with other Macs.

I am happy about doing this change. So, did you check that sysctl -n hw.perflevel0.logicalcpu works on Intel Macs?

from dotrun.

I am happy about doing this change. So, did you check that sysctl -n hw.perflevel0.logicalcpu works on Intel Macs?

Yes, this is working on Intel Macs (with latest macOS version), I can later check on older mac with older macOS to be sure (if we care about that).

from dotrun.

@jkfran so, it works on latest macOS (Monterey) both on M1 and Intel.

It does not work on old macOS (Mojave 10.14.6). It is couple years old OS and I don't know since when the hw.perflevel0.logicalcpu was introduced.

Maybe it would be possible to do sysctl -n hw.perflevel0.logicalcpu OR sysctl -n hw.ncpu? I'm not sure how to make conditionals in bash script in such way.

The other way would be not specify number of CPUs at all and let multipass use the default. Why we are not using the default? It is too small?

from dotrun.

TOTAL_CPUS=$(sysctl -n hw.perflevel0.logicalcpu || sysctl -n hw.ncpu) maybe?

This will still show error message when sysctl -n hw.perflevel0.logicalcpu fails I guess, but should still store correct value.

from dotrun.

@jkfran are you working on this issue?

from dotrun.

We discussed a v2 of dotrun based on Docker. Here are the meeting notes, so this issue shouldn't any longer apply.

I will notify everyone as soon as we get the first beta of dotrun v2.

from dotrun.

@jkfran are you actively working on this? If so, can you add estimate and epic please?

from dotrun.

I am closing this issue as it doesn't affect dotrun v2. Thank you!

from dotrun.