-------------------------- Docker Series-------------------------- ----------
Docker | Just read this article for docker in deep learning_Deep Learning Docker-CSDN Blog
Docker | Detailed steps to install portainer with docker-CSDN Blog
Docker | Push your own docker image to docker hub [picture and text details]-CSDN Blog
Docker is a tool for packaging, deploying, and running applications in lightweight containers . If you want to learn the basics of Docker, see the blog above.
1.Dockerfile writing rules
In the Dockerfile, everything on the left is INSTRUCTION, and the right is the ARGUMENT of these instructions .
Every instruction in the Dockerfile builds a layer, RUN and this is no exception. The behavior of each one RUN is the same as the process of manually establishing the mirror just now: create a new layer, execute these commands on it, and after the execution is completed, commit the modification of this layer constitutes a new mirror.
| Dockerfile instructions | explain |
|---|---|
| FROM | Specify base images that can be pulled from container registries (Docker hub, GCR, Quay, ECR, etc.) |
| RUN | Execute the command during the image build process. |
| ENV | Set environment variables in the image. It will be available during the build as well as in the running container. If you only want to set build-time variables, use the ARG directive. |
| COPY | Copy local files and directories to image |
| EXPOSE | Specify the port to be exposed for the Docker container. |
| ADD | It is a more feature-rich version of the COPY instruction. It also allows copying from a URL as source and automatically extracting tar files into images. However, it is recommended to use the COPY command instead of ADD. If you want to download a remote file, use curl or get it using RUN. |
| WORKDIR | Set the current working directory. You can reuse this directive in your Dockerfile to set different working directories. If WORKDIR is set, execute the , , , or instructions in the directory.RUNCMDADDCOPYENTRYPOINT |
| VOLUME | It is used to create volumes or mount volumes into Docker containers |
| USER | Set the username and UID when running the container. You can use this directive to set a non-root user for the container. |
| LABEL | Metadata information used to specify Docker images |
| ARG | Used to set build-time variables with keys and values. The ARG variable will not be available while the container is running. If you want to persist variables on a running container, use ENV. |
| SHELL | This directive sets shell options and the default shell for the RUN, CMD, and ENTRYPOINT directives that follow it. |
| CMD | It is used to execute commands in a running container. There can only be one CMD, if there are multiple CMDs, only the last one applies. It can be overridden from the Docker CLI . |
| ENTRYPOINT | Specifies the command that will be executed when the Docker container starts. If no ENTRYPOINT is specified, it defaults to ENTRYPOINT. You can also use the CLI to override ENTRYPOINT using flags . For more information, see CMD vs. ENTRYPOINT ./bin/sh -c--entrypoint |
Docker images are mounted sequentially (layer by layer) using federated mounting, and the image is built in the container.
2. Customize docker image
2.1. Method 1: Create an image through Dockerfile [most commonly used]
2.1.1.Writing Dockerfile
Create a new Dockerfile and write the environment
FROM nvcr.io/nvidia/pytorch:22.12-py3
RUN apt-get update \
&& apt-get install -y espeak
WORKDIR /app
COPY MB-iSTFT-VITS2 /app/MB-iSTFT-VITS2
WORKDIR /app/MB-iSTFT-VITS2
RUN /usr/bin/python -m pip install --upgrade pip \
&& pip install -r requirements.txt
EXPOSE 6843
ENV FLASK_APP=app.py
#CMD ["python", "/app/app.py"]
CMD ["flask", "run", "--host=0.0.0.0"]
2.2.2.Build the image
Build a Docker image using Dockerfile
The basic syntax is:
docker build [OPTIONS] PATH | URL | -docker build --no-cache -t tts:v0.1 -f ./Dockerfile .
2.3.3 Use of mirrors
View the build image
2.3.3.1. Use mirror directly
If the image is configured with flask or other network frameworks when encapsulating it, target recognition, speech synthesis, etc. can be directly implemented through customization. This article uses artificial intelligence algorithm model + flask framework!
docker run -p 5000:5000 tts:v0.1 python app.py --text
or
docker run -p 5000:5000 tts:v0.1 python app.py --text
If an error occurs, refer to [PS1]
2.3.3.2. Generate containers using encapsulated images
docker run -it --gpus all --ipc host --net host --name tts-test insanena414/tts_v1:latest /bin/bash
2.2. Method 2: Create through Git repo
docker build -t hello-world https://github.com/docker-library/hello-world.git#master:amd64/hello-worldThe command specifies the Git repo required for the build, and specifies the branch as and masterthe build directory as /amd64/hello-world/. Then Docker will go to git clone the project, switch to the specified branch, and enter the specified directory to start building.
2.3. Method 3: Create through tar compressed package
docker build my.tar.gzProblems encountered during the process and their solutions [PS]
【PS1】docker: Error response from daemon: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #1: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy'
Error analysis: It may be that the current cuda device version of the computer does not match the torch version.
references
【1】Use Dockerfile to customize the image · Docker -- from entry to practice (docker-practice.github.io)
【2】How To Build Docker Image [Comprehensive Beginners Guide] (devopscube.com)
【3】Quickly deploy frequently changing Docker images (feat. image size) | Miscellaneous blog (kimeuichan.github.io) 【4】Top 20 Dockerfile best practices for security – Sysdig