I've configured a ContainerApp with an auto scale rule in place based on the Azure Service Bus Topics. The up scaling of the ContainerApp works properly. However, it does not scale down. Any idea what is going wrong?
Replicas for ContainerApp:
Min replicas: 1
Max replicas: 5
Cooldown period: 300
Polling interval: 3
Auto scale rule for Container App:
Authentication
service-bus-connection-string: connection
Metadata
messageCount: 2
namespace: namespaceName
subscriptionName: subscriptionName
topicName: topicName
Update: One thing that I did notice is that there is a Probe Failed since the Health check is not configured in Aspire but I don't think that should be an issue since upscaling works properly.
I've configured a ContainerApp with an auto scale rule in place based on the Azure Service Bus Topics. The up scaling of the ContainerApp works properly. However, it does not scale down. Any idea what is going wrong?
Replicas for ContainerApp:
Min replicas: 1
Max replicas: 5
Cooldown period: 300
Polling interval: 3
Auto scale rule for Container App:
Authentication
service-bus-connection-string: connection
Metadata
messageCount: 2
namespace: namespaceName
subscriptionName: subscriptionName
topicName: topicName
Update: One thing that I did notice is that there is a Probe Failed since the Health check is not configured in Aspire but I don't think that should be an issue since upscaling works properly.
Share Improve this question edited Mar 28 at 7:19 NetDev asked Mar 28 at 7:19 NetDevNetDev 32 bronze badges 1- Ensure the health probe is properly configured, as a failed probe can prevent scaling down. Additionally, consider adjusting the cooldown period to be shorter (e.g., 60-120 seconds) to allow quicker scaling down based on message count @NetDev – Vinay B Commented Apr 2 at 8:32
1 Answer
Reset to default 0Container Apps scale down with scale rule doesnt work where as in place scale up works
Where your Azure Container App scales up correctly but fails to scale down is generally occurring due to scaling rule condition, cooldown configurations and misconfigurations among health probes.
In order to overcome this, you may need to reconfigure the health probe because When a container is marked unhealthy due to missing or failed probes, it can prevent to autoscale from scaling it down, even if the workload has decreased.
You can check out with the Microsoft documentation on configuring the health probe by following
refer: https://learn.microsoft/en-us/azure/container-apps/health-probes?tabs=arm-template
Once this done, check with the scaling rules dedicated to cooldown period and the messageCount threshold because you provided cooldown period is set to 300 seconds, which can delay scale-in significantly after a scale-out event. Try reducing this to around 60 or 120 seconds to allow the app to scale down more responsively.
Similarly, your messageCount threshold is set to 2 which is mostly a high value you can change it to 1 or even 0 temporarily to test whether scaling down is triggered as expected.
For more information on this refer
https://learn.microsoft/en-us/azure/container-apps/scale-app?pivots=azure-portal#configure-scaling-rules