Azure container instance is online, but sometimes not accessible

Question

I'm experiencing temporary unavailability with azure container instances. It's configured to be accesible through Front Door endpoint and origin group. I configured two container instances in origin group, so that one is always accessible when deploying changes.

Setup is successful, endpoint is available and working. After deploying, secondary container instance is turned off.

Occasionally it happens that endpoint stops being available. It returns "503 service unavailable" or "404". I can't give much details, since it happens randomly. Container instance is online and shows no logs and almost no cpu usage during this timeframe.
It starts working when waiting for some time. If I turn on secondary container instance it also starts workingI'd appreciate help with what's causing the issue. Or how could I find it.

Answer 1

Hi Aljaž Kraševec,

Thank you for reaching out to us on the Microsoft Q&A forum.

From what I understood, you are facing intermitent issues while working with ACI instances.

There are some situations that you may are experiencing limits on the requests, or limitation on Azure container services.

Please check the documentation below to better understand limits.

https://learn.microsoft.com/en-us/azure/container-instances/container-instances-resource-and-quota-limits.

Some other interesting docs - https://learn.microsoft.com/en-us/azure/reliability/reliability-containers.

Also please using Azure Log Analytics, to monitor your environement to identiy possible issues.

https://learn.microsoft.com/en-us/azure/container-instances/container-instances-log-analytics

Regarding availability, if you are looking for reliability and resilience when delivering your service. I would like to recommend to use AKS cluster (kubernetes), as ACI instances don't garantee high SLA and the infrastructure can change as needed, following Microsoft monitoring systems.

https://www.microsoft.com/licensing/docs/view/Service-Level-Agreements-SLA-for-Online-Services?lang=1

WORKAROUND

In the other hand, if you identify any issues with container instance being not stable, the recommendation is to STOP/START the ACI group, this will change the instances running in a specific cluster, moving the instances to another CLuster that are more stable and healty from monitoring perspective.

If the issue still happening, please reach out to our MS support team, opening a ticket and we will work on it.

If it was helpful, please click "Upvote or ACCEPT the Answer" on this post to let us know.

Thank You.

Lisboa