It looks like low priority VMs are getting their own quota. Is this true?

Question

Last week, we began to get errors while trying to scale up our VM Scale Sets that are configured to launch low priority VMs:

The operation couldn't be completed as it results in exceeding quota limit of LowPriorityCores. Maximum allowed: 10, Current in use: 14, Additional requested: 44.

This error message is interesting for a few reasons:

1. At the time we had around 20 VMs in the scale set, each with 2 vCPUs
2. It tells us that we were already over the quota
3. Our quota for that VM size is 200 vCPUs

While we were in this over-quota state, we were unable to create ANY VMs in that location -- not within that scale set and not regular, ad-hoc VMs.

This issue started in one location and has since spread to others without notice or warning.

Our only way to fix this was to delete the scale set and re-create it with regular VMs. We have since created a duplicate scale set with 5 VMs (10 vCPUs) which seems to be the limit of some new quota.

Since then, we've noticed that any low priority VMs we have running are not being counted against any of our vCPU quotas.

Is this the introduction of a new quota?

Answer 1

Following up on this, Reserved Instances now have their own quota separate from standard VM quota. The idea behind this is to provide a better customer experience as users can experiment and use the low priority VMs without impacting their overall cores that might be used for other uses. This also means customers can request additional low priority cores separate from standard cores which will increase flexibility of the cores available for your subscription. Currently all documentation where low priority cores can be used is being updated and should be completed within the next few weeks.