we have our Azure web apps and this had an issue yesterday .the node was low on memory and front end stopped .

Question

we have opted for Azure -Pay as you go plan .

Yesterday the warnign came : node was low on resoure memory .Threshold quantity 750Mi , available 703984Ki.Container was using 1584984Ki request is 500Mi, has larger consumtion of memory.

The application stopped and we had to increase the memory .

But with Pay as ypu go plan , should we have higher threshols ? or Azure will scale up to our requirements?

How do we address this issue.

WE did a major deployment yday and may be the applcation needs more resource .

How do we handle this .

Answer 1

With the Pay-As-You-Go plan, Azure provides the capability to scale resources based on your workload demands. You can define scaling rules and thresholds to trigger scale-out or scale-in actions. This means that Azure can automatically adjust the number of instances or resources according to your requirements, allowing for dynamic scaling based on actual demand. Therefore, having higher thresholds may not be necessary if Azure can scale up to meet your needs effectively.

To handle this type of scenarios , Azure provides autoscaling capabilities that allow for automatic scaling up and down based on various metrics. Azure Monitor autoscale can be configured to scale out resources when demand increases and scale in when demand decreases. This ensures that resources are allocated efficiently to meet performance requirements while optimizing costs.

You can set up scaling rules based on metrics such as CPU usage or memory usage, allowing the system to automatically adjust the number of instances as needed. For example, you can configure it to scale out by one instance if the average CPU/memory usage exceeds a certain threshold and scale in by one instance if the usage falls below a specified level.

References:

• Autoscaling
• Use Azure Monitor autoscale
• Configure autoscaling for an Azure solution

Please let me know if any further questions

Thanks

Deepanshu

Answer 2

It sounds like your application needs more memory after the recent deployment. Here’s a concise plan to address this:

1. Adjust Resource Limits: Increase the memory requests and limits for your containers in Kubernetes.
2. Monitor Usage: Use Azure Monitor to track memory usage and identify if the increase is temporary or ongoing.
3. Scale Appropriately: Consider scaling up (more resources per node) or scaling out (more nodes) using Azure Kubernetes Service (AKS).
4. Optimize Code: Check for memory leaks or inefficiencies in your application.

Azure’s Pay-As-You-Go plan can scale, but you need to configure it correctly.