When running a bash script during CodeBuild, I get this error:
./scripts/test.sh: line 95: docker: command not found
However, I've made sure to install docker at the start of the script using:
curl -sSL https://get.docker.com/ | sh
apt-get install -y docker-ce docker-compose
But this results in the following error:
Package docker-ce is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source
E: Package 'docker-ce' has no installation candidate
Any ideas on how to get docker working during CodeBuild?
There are a few different options for this in CodeBuild:
You can use CodeBuild provided images, which will already have docker installed on them. To use any one of these images select the privilege mode when creating the CodeBuild project.
You can enable Docker in custom image (images not managed by CodeBuild. e.g.: hosted in your ECR repo or public DockerHub) when configuring CodeBuild project. Select the privileged mode for your project settings. Instructions here: https://docs.aws.amazon.com/codebuild/latest/userguide/sample-docker-custom-image.html
Related
I am using AWS CodeBuild to build my application. I am using example build spec file as given here: https://docs.aws.amazon.com/codebuild/latest/userguide/build-spec-ref.html#build-spec-ref-example
I have already uploaded my custom Docker image to AWS ECR having requisites to build my application (Java/Scala based).
I get following error:
Reading package lists...
[Container] 2018/10/26 10:40:07 Running command echo Entered the install phase...
Entered the install phase...
[Container] 2018/10/26 10:40:07 Running command docker login -u AWS -p
.....
/codebuild/output/tmp/script.sh: 4: /codebuild/output/tmp/script.sh: docker: not found
Why should I get this error ? AWS CodeBuild is supposed to download this Docker image from ECR and then follow the instructions that I provide in the build spec file for building my application.
The example build.spec file assumes that your build image has Docker already installed. I was assuming "wrongly" that CodeBuild will install/configure Docker tools inside the image automatically.
Issue looks similar to AWS CodeBuild - docker: not found. I can't paste the same response to this question. So, please check my response there on how to enable Docker inside your build container to see if that solves your issue.
I'm not sure if this is Docker, the Elastick Beanstalk, or docker image issue related, but my problem is that I'm running the command eb local run to start the local environment alongside with docker.
Expected behavior
The command runs seamlessly
Actual behavior
ERROR: DockerVersionError - Your local host has the 'docker-py' version 1.10.6 Python package installed on it.
When you run a Multicontainer Docker application locally, the EB CLI requires the 'docker' Python package.
To fix this error:
Be sure that no applications on your local host require 'docker-py', and then run this command:
pip uninstall docker-py
The EB CLI will install 'docker' the next time you run it.
$ eb --version : EB CLI 3.12.2 (Python 2.7.1)
$ docker -v : Docker version 17.12.0-ce, build c97c6d6
If you want to launch multi-container Dockers using eb local run, you need to have uninstalled docker-py, and installed docker
As the error message indicates:
perform pip uninstall docker-py ** if you don't need it **.
run pip install "docker>=2.6.0,<2.7" immediately after
docker and docker-py cannot coexist. These release notes highlight the change in the package name. These release notes allude to the breakage the change in package name caused.
Not to be confused with Docker, the engine/client, docker-py/docker is a Python wrapper around the Docker client which the EBCLI relies on.
I'm using Elastic Beanstalk to deploy my application as a Single Docker Application.
My Dockerfile does composer install while deploying, but I get a Could not authenticate against github.com error.
I use these lines in my Dockerfile to install my dependencies:
WORKDIR /www
RUN ["composer", "install", "-o"]
How would I solve this issue?
I think you need to configure composer inside your container with your key or something like that, remember that inside your container you're basically on another os and you don't have public keys etc.
I'd try to install it from source rather than from git (as you don't have keys).
try this:
RUN curl -sS https://getcomposer.org/installer | php -- --install-dir=/usr/local/bin --filename=composer ()
I'm trying to deploy a docker application onto Elastic Beanstalk from Circle CI.
The deployment section of my circle.yml is
deployment:
hub:
branch: [internal, production]
commands:
- pip install awscli
- docker push company/web:$CIRCLE_SHA1
- sudo bash deploy.sh $CIRCLE_SHA1 $CIRCLE_BRANCH $CIRCLE_BUILD_NUM
and my deploy.sh calls aws cli as follows
aws --version
aws configure set aws_access_key_id $AWSKEY
aws configure set aws_secret_access_key $AWSSECRETKEY
aws configure set default.region us-west-2
aws configure set default.output json
echo "SAVING NEW DOCKERRUNFILE: $DOCKERRUN_FILE"
aws s3 cp $DOCKERRUN_FILE s3://$EB_BUCKET/$DOCKERRUN_FILE
But I get the error
--version: mispelled meta parameter?
sanity-check: "/root/.awssecret": file is missing. (Format: AccessKeyID\nSecretAccessKey\n)
configure: unknown command Usage: aws ACTION [--help]
The script works completely fine locally on mac os using the exact same key and secret.
Both versions (on circle and my mac) of awscli are 1.7.14
I'm Kevin from CircleCI. It looks like the issue here is related to the fact that when you install Python dependencies CircleCI installs them into a virtualenv. This is usually a great thing, as it isolates your python environment from the default system Python and supports our dependency cacheing. The problem here is that you're running your deploy.sh script with sudo, which clobbers the virtualenv environment and runs the default system version (which in this case is actually an older alternative AWS CLI). Dropping the sudo should fix things for you. (You would also be better off running pip install awscli==x.x.x in the "dependencies" phase, as it would be cached then.)
PS: Please contact sayhi#circleci.com for a timely response to questions in general.
As a starting point to making my own app that uses meanjs, I went to the meanjs website and used their yeomen generator to create the template/sample app. Following the instructions getting the sample application running out of the box on my local desktop machine worked within minutes. To complete the exercise I tried to deploy the sample app to an AWS/EC2 instance before making any changes to it. I have used the command line deployment tools in the past and liked it. Also it is nice how now you can just select an EC2 Linux instance with node and npm already installed and ready.
After checking the sample into git, I run "git aws.push" to deploy the app.
The problem is in the package.json the line:
"postinstall": "bower install --config.interactive=false"
In the eb-activity.log:
npm WARN cannot run in wd meansample#0.0.1 bower install --config.interactive=false (wd=/tmp/deployment/application)
The result is that AngularJS ends up not getting installed in /public/lib.
First thing I tried was giving the full path in the package.json file: node_modules/bower/bin/bower. This didn't help and results in the same error. Also noting that other commands like "grunt" don't need the full path specified in the package.json and they work.
I don't understand enough of the black box magic that aws.push does to understand why this error is happening. For example what user does it run as? What permissions does that user have? what options if any does it use when it runs npm install?
I did figure out a work-around, but it adds a lot of extra steps that shouldn't be required if aws.push was able to run bower install directly. Basically I can manually run the bower install in the ssh client connected to my EC2 instance, set the owner/group on the installed files, and restart the server.
Work-around steps:
1) On local command prompt run git aws.push. Wait for unsuccessfully deployment to finish.
2) Connect ssh client to EC2 instance. From the command prompt:
cd /var/app/current
/* NOTE: if I don't use sudo the ec2user I am logged in as does not have permission to create /public/lib needed to install AngularJS into*/
sudo node_modules/bower/bin/bower install --config.interactive=false --allow-root
/* NOTE: just changing the owner and group to match the same as the other files that aws.push deployed */
sudo chown -R nodejs public/lib
sudo chgrp -R nodejs public/lib
From AWS dashboard, select the correct EC2 instance, Action = Restart App Server(s)
Now AngularJS is install and the sample app works.
How do I eliminate the extra steps and make it so aws.push can do the bower install successfully?
I have experienced the same problem when trying to publish my nodejs app in a private server running CentOs using root user. The same error is fired by "postinstall": "./node_modules/bower/bin/bower install" in my package.json file so the only solution that was working for me is to use both options to avoid the error:
1: use --allow-root option for bower install command
"postinstall": "./node_modules/bower/bin/bower --allow-root install"
2: use --unsafe-perm option for npm install command
npm install --unsafe-perm