Auto scaling from zero capacity?

I’ve been trying to figure out how to scale from zero capacity without a master server, is this even possible?

Since you can’t make game sessions and player sessions if there’s no capacity, this means only full-admin credentials can add capacity? or is there something im missing here?


Hello @davevillz,

I don’t think I follow what you mean. If you have auto-scaling policies set up, GameLift would automatically scale up/down based on the rules that you have set up.

I hope that answers your question. Let us know if you need more information on how to set up auto-scaling policies.


Hey @davevillz, you’re correct that if you scale to zero capacity, there is currently no autoscaling rule which will be able to scale you back above zero. We currently recommend setting a minimum capacity of 1 to ensure you don’t autoscale to zero to prevent this issue from occurring.

The core problem is that there’s no metrics being emitted if there’s no capacity and no game sessions, hence there’s no data to autoscale on. I’d be interested in hearing more about how you’re thinking about this use case.

If you really do want to autoscale down to zero, the current workaround to get back above zero is to have a separate system like an AWS Lambda function, or maybe your matchmaking system have the credentials to scale up from zero when needed, then the GameLift autoscaling rules would kick back in again since they’ll have metrics data to scale on again.

1 Like

Thanks for the heads up on AWS Lambda, im not very familiar with it but sounds awesome, i was about to start coding my node.js master server but this should do it!

Im also more familiar with GameLift after i messed around with it for a month, i think a combination of a Lambda and a T2 fleet as a bare minimum while C3 fleets adds capacity would work in my use case (low player base with huge bursts)

Hello @Geoff

Is it still the case that Gamelift cannot autoscale from 0 instances? I have been trying to work with the QueueDepth metric thinking that the queue would be holding that information but it does not seem to be spinning up a new instance.

Sorry to necro, but I think this is important. At least one active instance incurs costs (got caught by this recently), so it would be nice if we could scale up from zero without an AWS lambda instance or something.

As for the QueueDepth metric, what I noticed is that there is two versions of it: one for the overall queue and one for the fleet. The rule-based metric goes off the fleet one, which for some reason remains at zero. This is what I mean: (dark blue is Queue depth and cyan is fleet Queue depth)

This is the configuration that I would like:

Ideally the process would be client calls StartGameSessionPlacement -> policy scale goes into effect -> add 1 new instance

Then, when active sessions drop to zero, we scale down until eventually zero instances (and no costs :slight_smile:)

1 Like

@Quinn - I know this has been on the GameLift backlog for a while to add as an option to scale up from Zero. I’ll let the service team know of your interest.

1 Like

Hi @Quinn,

With regards to your issue with QueueDepth metric: The queue depth metric for the fleet that is positioned first in the queue should match that of the queue itself. (see screenshot below).

If this is not the case, then please share your QueueArn / FleetArn and we can investigate what is going on.

Hi, sorry for the late response. I’m not sure I understand - what you have pictured looks like the queue depth of two different fleets to me.

This is what I see:

The queue depth metric for the fleet looks like it is above the queue depth of the queue itself