Why does the spin up time occasionally vary on the macOS executor?

Slightly longer spin up times can occasionally occur with the macOS executor due to the way pre-scaling works. CircleCI automatically spins up a certain number of macOS VMs per image/resource class combination and the exact number varies depending on the demand we are seeing. Pre-scaling is always adjusting based on current demand and capacity availability, but it is not always able to predict spikes in demand.

If your job request is able to be fulfilled by a VM that is already booted up and ready to go, then you will see a very short spin up time - usually less than 30 seconds.

If, however, we do not have any VMs available at the time of the request, then we need to spin up a new VM and this can take up to a few minutes. Generally this is more likely to happen if you are using a less popular image/resource class combination as CircleCI will be pre-booting less of these.

This is normal behaviour and is something we constantly monitor closely. We cannot guarantee that there are enough pre-booted VMs for every job, but we do try to ensure this is the case as much as we can to provide a better customer experience.

If you find your jobs are consistently taking an excessive amount of time to spin up (more than five minutes), then please reach out to us on a support ticket so we can look into this further.

Was this article helpful?
0 out of 0 found this helpful

Comments

0 comments

Article is closed for comments.