Elastic Container Instance:Scale up the temporary storage space

How it works

The 30 GiB default is shared between the container image and your workload. When ECI starts a pod, it pulls and decompresses the container image into temporary storage. The decompressed image typically occupies at least twice the compressed image size — the exact amount depends on the compression ratio. To estimate how much space your workload actually has available, subtract the decompressed image size from the total allocated capacity:

Usable space = total allocated capacity - decompressed image size

For example, if you allocate 50 GiB and your image decompresses to 8 GiB, your workload has approximately 42 GiB available.

Choose a method

Two methods are available. Each uses different Kubernetes semantics:

Limit vs. request semantics: The annotation sets a hard limit — the pod receives exactly the capacity you specify. The resource request influences scheduling decisions, and the actual allocated capacity may differ from the value you specify depending on the image cache type. For details, see Image cache type and storage capacity.

If you configure both methods, the greater value takes effect.

Prerequisites

Before you begin, ensure that you have:

• A Kubernetes cluster with ECI as the container runtime

• Permission to create or manage pods in your cluster

Scale up using an annotation

Add the k8s.aliyun.com/eci-extra-ephemeral-storage annotation to the pod's metadata section. Set the value to the amount of additional capacity you want — not the total.

The following example adds 20 GiB of additional capacity, giving the pod 50 GiB in total:

apiVersion: v1
kind: Pod
metadata:
  name: test
  annotations:
    k8s.aliyun.com/eci-extra-ephemeral-storage: "20Gi"  # Additional capacity on top of the default 30 GiB
spec:
  containers:
  - name: nginx
    image: registry-vpc.cn-beijing.aliyuncs.com/eci_open/nginx:1.14.2
    imagePullPolicy: IfNotPresent
  restartPolicy: Always

Scale up using a resource request

Set the ephemeral-storage field under resources.requests to the total capacity you want. The value must exceed 30 GiB — values at or below 30 GiB have no effect. The additional capacity equals the value you set minus 30 GiB.

The following example sets a total of 50 GiB, resulting in 20 GiB of additional capacity:

apiVersion: v1
kind: Pod
metadata:
  name: test
spec:
  containers:
  - name: nginx
    image: registry-vpc.cn-beijing.aliyuncs.com/eci_open/nginx:1.14.2
    imagePullPolicy: IfNotPresent
    resources:
      requests:
        ephemeral-storage: 50Gi  # Total capacity: 30 GiB default + 20 GiB additional
  restartPolicy: Always

Apply and verify

1. Save your configuration to a file (for example, extra-storage.yaml) and create the pod:

   kubectl apply -f extra-storage.yaml

2. Confirm the annotation value:

   kubectl describe pod test | grep k8s.aliyun.com/eci-extra-ephemeral-storage

   The following output confirms that 20 GiB of additional capacity was applied:

   k8s.aliyun.com/eci-extra-ephemeral-storage: 20Gi

3. Log in to the container and check the file system:

   kubectl exec -it test -- sh
   df -h

   In the output, the overlay line shows the total available storage. A value of 50G confirms that the scale-up is in effect:

   Filesystem      Size  Used Avail Use% Mounted on
   overlay          50G  2.1G   48G   5% /
   tmpfs            64M     0   64M   0% /dev
   tmpfs           1.9G     0  1.9G   0% /sys/fs/cgroup
   /dev/vda4        50G  2.1G   48G   5% /etc/hosts
   shm              64M     0   64M   0% /dev/shm
   tmpfs           4.0G   12K  4.0G   1% /run/secrets/kubernetes.io/serviceaccount
   tmpfs           1.9G     0  1.9G   0% /proc/acpi
   tmpfs           1.9G     0  1.9G   0% /sys/firmware

Image cache type and storage capacity

The actual storage layout depends on the image cache type used when the pod is created.

Automatic image cache: The temporary storage space resides on the system disk. The capacity of the system disk partition (/dev/vda5) equals the total temporary storage capacity.

Manual image cache: The temporary storage space resides on a data disk. The system disk partition (/dev/vda5) remains at the default 30 GiB and is not used for temporary storage.

This difference explains why the actual scaled-up capacity from a resource request may differ from the value you specify.

What's next

• Billing of the temporary storage space