ACI Instances Unable to Pull Images from ACR

Thanks Rashmika for the explanation. I tried the File Share option, but the instance still fails to pull the image multiple times in the Events before it eventually succeeds. This is the command for creating one container group with one instance

az container create \
  --resource-group rg-container-se14-tembi-eip-prod \
  --name aci-se14-tembi-checkout-hf-prod0 \
  --os-type Linux \
  --image crtembieipprod.azurecr.io/checkout-scraper-hf:latest  \
  --azure-file-volume-share-name aci-se11-tembi-checkout-hf-prod0-fileshare \
  --azure-file-volume-account-name strgse11checkouthfprod0 \
  --azure-file-volume-account-key UBthHmtVR3SoLRquKu7S4knA0oWs8LP2ZYInWGtICPyiTZ8Qtw9YP8jtw6dUaghZEx6fe4W9wCDQ+ASt2gADjg== \
  --azure-file-volume-mount-path /mnt/storage \
  --memory 2 \
  --cpu 0.5 \
  --subnet subnet-scraper \
  --vnet vnet-se14-tembi-scraper-eip-prod

Hi Moustafa Kotb,

Thanks for following the steps and sharing the error. The error you encounter is still related to disk space. The reason the Azure File Share isn't fixing this is because it doesn't solve the problem at the right stage of the process.

The "not enough space on the disk" error happens during the image pull and unpack process, which occurs on the ACI host's temporary (ephemeral) storage. This step happens before your container starts and the Azure File Share is mounted. The file share provides runtime storage for your application, not the initial staging area for the container image.

ACI hosts have a fixed, small amount of temporary disk space (around 30 GB). When your large image is uncompressed, it temporarily exceeds this limit, causing the deployment to fail. The occasional success is due to the instance being scheduled on a host that, by chance, has slightly more free space at that moment.

To resolve this, you can either reduce image size or use a different service

1. Reduce Image Size: The most effective solution is to optimize your image using multi-stage Docker builds and a smaller base image (e.g., alpine or distroless). This directly addresses the root cause of the problem.
2. Use a Different Service: If a smaller image is not an option, consider migrating to an Azure service designed for larger workloads:
   • Azure Container Apps: A fully managed service that can support larger images.
     • Azure Kubernetes Service (AKS): Provides full control over your nodes, allowing you to configure VMs with larger OS disks to accommodate your images consistently.

The Azure File Share is a tool for providing persistent storage for your running application, but it cannot be used to stage the container image itself.

For reference: Troubleshoot common issues - Azure Container Instances | Microsoft Learn

Resource availability & quota limits for Azure Container Instances (ACI) - Azure Container Instances | Microsoft Learn

Let me know if you require any additional assistance from my end. I am happy to help you with the queries.

Thanks,

Rashmika

Hi Moustafa Kotb,

Just following up to check whether you were able to check the above steps and references provided to resolve your issue. Let me know if you require any additional assistance from my end. I am happy to help you with the queries. If the information is helpful, please click on Upvote and Accept Answer on it so that it can help other community members.

Thanks,

Rashmika