Elastic Container Instance:Usage scenarios

Background information

If you want to build cluster applications, we recommend that you use Container Service for Kubernetes (ACK) clusters or ACK Serverless clusters to fully utilize Elastic Container Instance. If your business is small or you want to deploy only a few standalone applications, ECI-Client allows you to use Elastic Container Instance more easily.

ECI-Client uses the same syntax to manage elastic container instances as Docker. If you are familiar with the Docker CLI, you can quickly get started with Elastic Container Instance. This topic describes common usage scenarios of ECI-Client.

Test container applications

If you are developing an application based on Elastic Container Instance, you may need to frequently deploy elastic container instances to check whether the applications run as expected. In this case, you can use ECI-Client to conveniently deploy elastic container instances.

For example, you build a centos:8.1.1911 image and upload the image to a Docker Hub image repository. To test whether the image works as expected, you can run the following command in ECI-Client to deploy an elastic container instance:

eci run -w 5 centos:8.1.1911 sleep 3600

After you start an elastic container instance, you can run the exec command to log on to containers in the instance. If the instance ID is eci-uf60grb03kz2nlm2****, run the following command:

eci exec -ti eci-uf60grb03kz2nlm2**** bash

Train a simple model

In the following example, a sample image that contains a Python program is used. After you start the container, the program downloads data from the Internet and uses TensorFlow to train the data. The trained data is stored in the /tmp/imagenet directory.

eci run --name test-tensor \
--type ecs.gn6i-c4g1.xlarge \
--gpu 1 \
-w 5 \
--volume /tmp/imagenet \
registry.cn-hangzhou.aliyuncs.com/eci_open/tensorflow:1.0 \
python /home/classify_image/classify_image.py

TensorFlow programs rely on GPUs. You must use the --type parameter to specify a GPU-accelerated instance type of Elastic Compute Service (ECS) for the elastic container instance. You must also use the --gpu parameter to allocate GPUs to containers in the instance, and use the --volume parameter to mount a disk to the elastic container instance. The disk stores the training results.

After you start the elastic container instance, you can run the eci logs command to view the log output of TensorFlow programs. If the instance ID is eci-uf60grb03kz2nlm2****, run the following command:

eci logs eci-uf60grb03kz2nlm2****

Deploy a personal website

In the following example, ECI-Client is used to deploy a WordPress website that uses a MySQL database. The sample image is from a DockerHub image repository:

• WordPress: wordpress:4.9.8-php5.6-apache

• MySQL: mysql:5.7

1. Deploy the MySQL database and configure a password for the root user.

   eci run -w 5 --name mysql -e MYSQL_ROOT_PASSWORD=passwd*** -tid mysql:5.7

   After you run the command, an instance ID is generated. You can run the eci ps command to check the instance status. If the instance status changes to Running, the MySQL database is deployed. The following output is returned:

   CG_ID                      FIRST_IMAGE   COMMAND     CREATED                     STATUS    PUBLIC_IP        NAME     
   eci-2zecsa60voctfzr2****   mysql:5.7                 2023-02-17T16:33:22+08:00   Running   101.200.XX.XX    mysql  

2. Obtain the internal IP address of the MySQL database.

   In the following example, a jq command is run to obtain the internal IP address of the MySQL database. If no jq command is available on the on-premises device, you must first install the jq command sets.

   eci inspect eci-2zecsa60voctfzr2**** | jq -r .ContainerGroups[0].IntranetIp

   Expected output:

   172.16.XX.XX

3. Deploy WordPress.

   You can use the environment variable -e to pass in the address, username, and password of the MySQL database. You can also use -w to assign a public IP address to the WordPress website and set the bandwidth of the website to 5 MB.

   eci run --name wordpress \
   -e WORDPRESS_DB_HOST=172.16.XX.XX \
   -e WORDPRESS_DB_USER=root \
   -e WORDPRESS_DB_PASSWORD=passwd*** \
   -w 5 \
   -tid wordpress:4.9.8-php5.6-apache

   After you run the command, an instance ID is generated. You can run the eci ps command to check the instance status. If the instance status changes to Running, the WordPress website is deployed. The following output is returned:

   CG_ID                      FIRST_IMAGE                     COMMAND   CREATED                     STATUS    PUBLIC_IP        NAME     
   eci-2zeedp8bxor356b7****   wordpress:4.9.8-php5.6-apache             2023-02-17T16:39:51+08:00   Running   101.200.XX.XX    wordpress
   eci-2zecsa60voctfzr2****   mysql:5.7                                 2023-02-17T16:33:22+08:00   Running   101.200.XX.XX    mysql

   You can view the public IP address of the WordPress website in the output. In this example, the public IP address is 101.200.XX.XX. Enter http://100.100.XX.XX in your browser to access the WordPress website.