New Scaling tab (with Autoscaling)

We are doing a phased roll out of the Scaling tab (and autoscaling) to all Render customers. It includes the ability to turn on CPU-based autoscaling for all eligible services, with memory-based autoscaling in the works.

We’ve also moved the Instances setting from the Settings tab to the Scaling tab; it is now available as Manual Scaling.

If you have any questions please post in the community or email Thank you!


How do I set the auto-scaling min, max instances using YAML?

Render YAML hasn’t supported autoscaling configuration. We are working on it and the ETA is the end of this Q2.

As part of the autoscaling beta i had no issues using it. Right now, there is no change to the number of instances no matter how i configure it.

I am looking into it.

@Peter_Schroder We have found the root cause but we are still working on the solution. I will keep you posted. Sorry for the inconvenience.

1 Like

works now.

1 Like

Thanks for your confirmation!
We have deployed the fix. We will keep ramping this feature.

1 Like

We are having issues with Autoscaling? or general availability in our web-instances in frankfurt. The status dashboard indicates just errors with new deploys. There is also no way to run a console on the instances. Looks like some general availability issue.

There is an ongoing incident affecting EU cluster. We are working on a solution. Sorry for the inconvenience.

yeah, the status page was updated after posting this. thank’s for the feedback though

Memory-based autoscaling is on production! You can toggle and specify the target CPU and memory utilization separately. Please check this document for more information.

If you have any questions please post in the community or email Thank you!


Hey @Hao-Ji_Wu!

I setup autoscaling for my web service which has no users yet. To make sure everything works well, I load tested the service and the number of instances scaled up according to my target CPU and memory utilization of 50%

It’s been 1 hour since the load test and the memory utilization is now stable at 30% and the CPU utilization is at 0%.

But I can see that 2 instances of the web service are active while I expect only 1 instance to handle the load at this point in time.

Could you please look into the issue? Thank you!

The desired number of instances is calculated by the ratio of the current utilization and target utilization. In your case, after scaling up 2 instances and memory utilization is to 30%, the desired number of instances will be ceil[2 * 30%/50%] = 2 (50% is the target memory utilization). Therefore the number of instances is unchanged.

You can refer to this doc for more information. If you need further investigation please DM your service name or id to me. Thanks.

1 Like