All Products
Search
Document Center

Elastic Container Instance:Release notes

Last Updated:Oct 18, 2024

This topic describes the release notes for Elastic Container Instance since 2023 and provides links to the relevant references.

Note

2024

July 2024

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Forced deletion of elastic container instances

When you delete an elastic container instance, you can forcibly delete resources without waiting for the timeout period of graceful termination.

2024-07-01

DeleteContainerGroup

N/A

Setting of a maximum pending duration

By default, the maximum pending duration of a pod is 4 hours. You can specify the maximum pending duration based on your business requirements. The system automatically terminates the pod after the pod times out. This helps you prevent issues that are caused by the failure to handle abnormal pods in a timely manner.

2024-07-01

Not supported

Specify the maximum pending duration of an Elastic Container Instance-based pod

June 2024

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Custom DNS server

In hybrid cloud scenarios, if you want to pull images from an image repository in an internal network or a container needs to access an internal domain name, you can set an internal Domain Name System (DNS) server for the Elastic Container Instance-based pod (pod) to resolve the internal domain name.

2024-06-28

Not supported

Configure a custom DNS server for a pod

Using tags in the management of Elastic Container Instance resources

The TagResources, ListTagResources, and UntagResources API operations are published to add tags to Elastic Container Elastic resources or remove tags from the resources.

2024-06-20

Tags

N/A

Support for the eighth generation Elastic Compute Service (ECS) instance families in the filtering of ECS instance families and generations of ECS instance families

You can specify or exclude the eighth generation ECS instance families when you create pods by specifying or excluding ECS instance families or specific generations of ECS instance families.

2024-06-20

Not supported

Stopping the support for the ecs.t6 and ecs.t5 ECS instance families

When you specify ECS instance types to create an elastic container instance, the ecs.t6 and ecs.t5 burstable instance families are no longer supported.

2024-06-04

Specify x86-based ECS instance types to create an elastic container instance

ECS instance types of the x86 architecture

May 2024

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Support for the general compute category for creation of elastic container instances and pods

When you specify vCPU and memory specifications to create an elastic container instance, you can further specify a compute category. If your business has high performance requirements, you can select the general compute category.

2024-05-31

Specify a compute category to create an elastic container instance

Specify a compute category to create a pod

Support for CloudMonitor metrics such as metrics about GPUs and volumes

Metrics about GPU and volumes are supported when you use CloudMonitor to monitor elastic container isntances.

2024-05-23

Use CloudMonitor to monitor elastic container instances

Use CloudMonitor to monitor elastic container instances

April 2024

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Support for modification of kernel.shm_rmid_forced in security contexts

You can modify the sysctl parameter kernel.shm_rmid_forced when you configure security contexts.

2024-04-23

Configure a security context for an elastic container instance or a container

Configure a security context for a pod or container

Switch of GPU driver versions

You can switch the GPU driver version when you call API operations to create an GPU-accelerated elastic container instance.

2024-04-15

Specify GPU-accelerated ECS instance types to create an elastic container instance

N/A. Supported in earlier release notes.

Support for the g5ne ECS instance family in the filtering of ECS instance families and generations of ECS instance families

You can specify or exclude the g5ne ECS instance family when you create pods by specifying or excluding ECS instance families or specific generations of ECS instance families.

2024-04-08

Not supported

January 2024

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Support for the ecs.c8i, ecs.c8a, and ecs.g8i ECS instance families

The ecs.c8i, ecs.c8a, and ecs.g8i ECS instance families are supported when you create elastic container instances by specifying ECS instance types.

2024-01-25

Specify x86-based ECS instance types to create an elastic container instance

ECS instance types of the x86 architecture

2023

December 2023

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Support for the economy compute category for creation of elastic container instances

If you specify vCPU and memory specifications to create an elastic container instance and the number of vCPUs of the instance is from 0.25 to 8, you can further specify the economy compute category for the instance to reduce the cost of the instance.

2023-12-29

Specify a compute category to create an elastic container instance

Specify a compute category to create a pod

Ignored NotReady state of the sidecar container

If you use a sidecar container to implement an effect that is similar to the effect of DaemonSets and the status of the sidecar container is NotReady, the status of the pod is NotReady. In specific scenarios, if you do not want the status of the sidecar container to affect the status of the entire pod, you can use environment variables to ignore the status of the sidecar container.

2023-12-29

N/A

Ignore the NotReady state of the sidecar container

Creation of pods after Cluster IP address synchronization

In a Container Service for Kubernetes (ACK) cluster, Elastic Container Instance-based pods and the component that synchronizes Services are started and run in parallel. If the cluster contains a large number of Services and Service endpoints, the access to the pods may time out because the cluster IP address is not synchronized among the Services.

To resolve this issue, Elastic Container Instance allows the system not to create a pod until the cluster IP address is synchronized among Services. This ensures that containers in the pod can access Services.

2023-12-29

N/A

Create a pod after the cluster IP address is synchronized

Additional specifications for elastic container instances

If you create elastic container instances by specifying the number of vCPUs and memory size, you can select specifications including 2 to 8 vCPUs and memory sizes with a small step size for some regions.

2023-12-27

Specify the number of vCPUs and memory size to create an elastic container instance

Specify the number of vCPUs and memory size to create a pod

November 2023

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Force termination of the sidecar container and ignored container exit code

If you use a sidecar container to implement an effect that is similar to the effect of DaemonSets, a job-type pod may not be able to complete running. In this case, you can configure environment variables to specify the type of the sidecar container and ignore the exit code of the sidecar container. This ensures that the job-type pod can complete running as expected.

2023-11-30

N/A

Forcibly terminate the sidecar container and ignore the container exit code

October 2023

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Support for the ecs.gn7s instance family

The ecs.gn7s instance family is supported if you specify ECS GPU-accelerated instance types to create GPU-accelerated elastic container instances.

2023-10-17

Specify GPU-accelerated ECS instance types to create an elastic container instance

GPU-accelerated ECS instance types

Encryption of temporary storage space

Elastic Container Instance allows you to encrypt the temporary storage space of an elastic container instance to protect sensitive image or business data or to comply with regulations. This way, your data stays secure and intact against unauthorized access or leakage.

2023-10-16

Not supported

Encrypt the temporary storage space

September 2023

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Support for specific ECS instance types

Specific ECS instance types, such as GPU-accelerated ECS instances types and ECS instance types that provide high clock speeds, are supported. You can select these instance types when you create elastic container instances in the Elastic Container Instance console and your business requires special specifications.

2023-09-13

Support for specific Elastic Compute Service (ECS) instance types

N/A

August 2023

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Configuration of the startup and exit priorities of containers

Environmental variables can be added to containers in elastic container instances to ensure that the containers start or exit in sequence.

2023-08-22

Configure the startup and exit priorities of containers

Configure the startup and exit priorities of containers

Support for DataCaches

This feature allows you to create a cache of a large amount of business data, such as model training data. When you create an elastic container instance, you can mount and use the cached data. This reduces the startup time of the instance, prevents repeated data downloads, and reduces the costs of using the instance.

2023-08-21

Overview

Overview

Support for the ecs.u1 instance family in the filtering of ECS instance families and generations of ECS instance families

You can specify or exclude the ecs.u1 ECS instance family when you create pods by specifying or excluding ECS instance families or specific generations of ECS instance families.

2023-08-18

Not supported

Support for private IP addresses

A private IP address can be specified when you create an elastic container instance.

2023-08-14

Specify a private IP address for an elastic container instance

Specify a private IP address for a pod

Modification of the vm.min_free_kbytes parameter

The vm.min_free_kbytes parameter can be modified when you modify sysctl parameters by configuring a security context.

Note

We recommend that you specify a value for vm.min_free_kbytes that is not greater than 20% of the total memory size.

2023-08-10

Configure a security context for an elastic container instance or a container

Configure a security context for a pod or container

Use of CloudMonitor to monitor elastic container instances

CloudMonitor can be used to monitor elastic container instances.

2023-08-04

Use CloudMonitor to monitor elastic container instances

Use CloudMonitor to monitor elastic container instances

Cross-zone distribution of Elastic Container Instance-based pods and configuration of affinities

In ACK Pro clusters or ACK Serverless Pro clusters, Elastic Container Instance-based pods can be distributed across zones based on Kubernetes native scheduling semantics to ensure the high availability of the pods. In addition, affinities can be configured to deploy pods in specific zones based on Kubernetes native scheduling semantics to ensure the high performance of the pods.

2023-08-03

Not supported

Spread Elastic Container Instance-based pods across zones and configure affinities

July 2023

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Image pulling from Container Registry instances across Alibaba Cloud accounts without using a Secret

If a Container Registry instance and the elastic container instance that you want to create do not belong to the same Alibaba Cloud account, Resource Access Management (RAM) users can be used to pull images from the Container Registry instance without using a Secret.

2023-07-19

Pull images from a Container Registry instance without using a secret

Pull images from a Container Registry instance without using a secret

Refined collection of stdout logs

The following values are supported for the collection path of stdout logs:

  • stdout: collects stdout logs and standard error (stderr) logs.

  • stderr-only: collects stderr logs, but does not collect stdout logs.

  • stdout-only: collects stdout logs, but does not collect stderr logs.

2023-07-18

Configure log collection for an elastic container instance

Collect logs by using environment variables

Modification of the net.ipv4.ip_local_port_range parameter

The net.ipv4.ip_local_port_range parameter can be modified when you modify sysctl parameters by configuring a security context.

2023-07-04

Configure a security context for an elastic container instance or a container

Configure a security context for a pod or container

June 2023

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Support for the specification of GPU driver versions

In some scenarios, you may need to use different driver and CUDA versions for different GPU-accelerated elastic container instances. In this case, you can add annotations to specify the driver and CUDA versions.

2023-06-30

Not supported

Specify ECS instance types to create a pod

Support for the ecs.u1 instance family

The ecs.u1 instance family can be used when you create elastic container instances by specifying ECS instance types.

2023-06-20

Specify ECS instance types to create an elastic container instance

Specify ECS instance types to create a pod

Configuration of fixed IP addresses

A fixed IP address can be configured to create an elastic container instance. After the instance is released, the system retains the fixed IP address. If you create an elastic container instance that meets specific conditions during the validity period (48 hours by default), the system assigns the retained IP address to the instance.

2023-06-09

Configure an elastic container instance to use a fixed IP address

Configure an Elastic Container Instance-based pod to use a fixed IP address

Support for multiple security groups

Multiple security groups can be specified by using the k8s.aliyun.com/eci-security-group annotation. The following requirements must be met:

  • You can assign up to five security groups.

  • The assigned security groups must belong to the same virtual private cloud (VPC).

  • The assigned security groups must be of the same type.

2023-06-05

Not supported

Assign security groups to a pod

May 2023

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Support for elasticity assurances

Elasticity assurances are supported to provide guaranteed access to resources that you can use to create pay-as-you-go elastic container instances. For example, you must use a specific number of pods of a specific ECS instance type for a period of time. In this case, we recommend that you purchase an elasticity assurance to obtain reserved resources. This prevents instance creation failures due to insufficient resources and negative impacts on your business.

2023-05-17

Not supported

Use the private pool of an elasticity assurance to create a pod

Additional vCPU and memory specifications

The following specifications are added:

  • 24 vCPUs and 24 GiB of memory

  • 32 vCPUs and 32 GiB of memory

2023-05-11

Specify the number of vCPUs and memory size to create an elastic container instance

Specify the number of vCPUs and memory size to create a pod

Automatic creation of pay-as-you-go elastic container instances

The k8s.aliyun.com/eci-spot-fallback: "true" annotation can be added when you create a preemptible elastic container instance. If the resources that meet the requirements for the specification of the preemptible instance are insufficient, a pay-as-you-go instance is automatically created.

2023-05-09

Not supported

Create a preemptible elastic container instance

March 2023

Feature

Description

Release date

References

(Elastic Container Instance console and API)

References

(Kubernetes)

Support for clusterDNS, clusterDomain, and customResources in VNodectl

customResources, clusterDNS, and clusterDomain can be configured by using parameters when you create or update VNodes.

2023-03-28

N/A

This feature is applicable only to self-managed Kubernetes clusters. For more information, see Common commands for VNodectl.

Filtering of specific ECS instance families

When you create an Elastic Container Instance-based pod by specifying vCPU and memory specifications, the system uses a variety of ECS instance types to support the creation and automatically selects the instance types that meet the vCPUs and memory requirements and have sufficient resources. The k8s.aliyun.com/eci-instance-family annotation can be added to include or exclude specific ECS instance families when you create a pod by specifying the number of vCPUs and a memory size.

2023-03-07

Not supported

Specify or exclude specific ECS instance families to create a pod

Filtering of specific generations of ECS instance families

When you create an Elastic Container Instance-based pod by specifying vCPU and memory specifications, the system uses a variety of ECS instance types to support the creation and automatically selects the instance types that meet the vCPUs and memory requirements and have sufficient resources. The k8s.aliyun.com/eci-instance-generation annotation can be added to include or exclude specific generations of ECS instance families when you create a pod by specifying the number of vCPUs and a memory size.

2023-03-07

Not supported

Specify or exclude specific generations of ECS instance families to create a pod

Configuration of custom hosts

In some scenarios, you may need to configure hosts for an Elastic Container Instance pod. For example, when you pull an image from a self-managed image repository, you must specify the actual IP address of the image repository by using the hosts file. The k8s.aliyun.com/eci-custom-hosts annotation can be added to specify the hosts of pods.

2023-03-03

Not supported

Configure hosts for an Elastic Container Instance-based pod

Support for Arm-based instances

Arm-based ECS instance types can be specified to create Arm-based elastic container instances.

Arm-based ECS instance types is based on the Arm architecture. Each vCPU corresponds to a physical core of a processor. The Arm architecture delivers stable performance by using dedicated resources.

2023-03-03

Create an Arm-based instance