Questions[travis-ci](server)

I am successfully building the redisearch.so module on travis-ci.

After the module is built, I instruct the Redis to load it like this

redis-cli MODULE LOAD $TRAVIS_BUILD_DIR/rs/build/redisearch.so

which gives me the following error

5439:M 17 Nov 2020 09:12:55.341 # Module /home/travis/build/***/rs/build/redisearch.so failed to load: /home/travis/build/***/rs/build/redisearch.so: cannot open shared object file: Permission denied

this is what ls -la /home/travis/build/***/rs/build/redisearch.so gives me

-rwxrwxr-x 1 travis travis 2711408 Nov 17 09:12 /home/travis/build/***/rs/build/redisearch.so

The question is: how do I fix this permission denied error?

PS:

I have already tried to chmod and chown the /home/travis/build/***/rs/build/redisearch.so but with no luck.

UPD: I have moved the module to /etc and it just worked

I have a PHP project on GitHub for which I would like to build versioned Docker images. Kinda like CoreOS Container Linux does with their alpha, beta and stable release channels.

For development I follow the Git Flow principle and use the feature -> develop -> master branching.

I have hooked up Travis for automated PHPUnit testing and added CircleCI today as I found a Rakefile which could automatically generate a version. Which is used as a tag.

My current .travis.yml:

---
services:
  - docker
sudo: required
env:
  global:
    - ...

addons:
  jwt:
    secure: ...
cache:
  directories:
    - /home/travis/docker/
before_install:
  - "docker --version"
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then echo "ENV GIT_SHA ${TRAVIS_COMMIT::8}" >> Dockerfile; fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]] && [[ -f ${DOCKER_CACHE_FILE} ]]; then gunzip -c ${DOCKER_CACHE_FILE} | docker load; fi'
before_script:
  - "env > .env"
install:
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then docker build -t ${DOCKER_REPOSITORY}:${TRAVIS_COMMIT::8} --pull=true .; fi'
language: php
notifications:
  slack:
    secure: ...
php:
  - "5.6"
script:
  - "phpunit --bootstrap tests/bootstrap.php --testdox tests --coverage-text"
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then mkdir -p $(dirname ${DOCKER_CACHE_FILE}) ; docker save $(docker history -q ${DOCKER_REPOSITORY}:${TRAVIS_COMMIT::8} | grep -v "<missing>") | gzip > ${DOCKER_CACHE_FILE}; fi'
after_success:
  - 'if [[ $TRAVIS_PHP_VERSION == "5.6" ]] && [[ $TRAVIS_BRANCH == "develop" ]] && [[ $TRAVIS_PULL_REQUEST == "false" ]]; then sh generate-api.sh; fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then docker login -u ${DOCKER_USERNAME} -p ${DOCKER_PASSWORD}; fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then docker tag ${DOCKER_REPOSITORY}:${TRAVIS_COMMIT::8} ${DOCKER_REPOSITORY}:${TRAVIS_COMMIT::8}; fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then docker tag ${DOCKER_REPOSITORY}:${TRAVIS_COMMIT::8} ${DOCKER_REPOSITORY}:travis-${TRAVIS_BUILD_NUMBER}; fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then docker tag ${DOCKER_REPOSITORY}:${TRAVIS_COMMIT::8} ${AWS_ACCOUNT_NUMBER}.dkr.ecr.eu-west-1.amazonaws.com/${ECR_REPOSITORY}:${TRAVIS_COMMIT::8}; fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then docker push ${DOCKER_REPOSITORY}; fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then pip install --user awscli; fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then export PATH=$PATH:$HOME/.local/bin; fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then eval $(aws ecr get-login --region eu-west-1); fi'
  - 'if [[ ${TRAVIS_BRANCH} == "master" ]] && [[ ${TRAVIS_PULL_REQUEST} == "false" ]]; then docker push ${AWS_ACCOUNT_NUMBER}.dkr.ecr.eu-west-1.amazonaws.com/${ECR_REPOSITORY}:${TRAVIS_COMMIT::8}; fi'

And my current circle.yml:

---
machine:
  services:
    - docker
  php:
    version: 5.6.22

dependencies:
  override:
    - docker info
    - gem install httparty
    - rake build

test:
  override:
    - mkdir -p $CIRCLE_TEST_REPORTS/phpunit
    - phpunit --bootstrap tests/bootstrap.php --log-junit $CIRCLE_TEST_REPORTS/phpunit/junit.xml tests
    - docker run -d -p 9000:9000 storecore/php:latest; sleep 10
    - nc -z -w5 localhost 9000

And the used Rakefile:

require 'rake'
require 'httparty'
require 'json'

# Read the base version from VERSION file.
def version
  file = File.readlines('./version.php')
  v = file[1].scan(/'([^']*)'/)
  v[1].join(",")
end

# The name of the container
def container_name
  'php'
end

# The username the container is pushed to on DockerHub
def username
  'storecore'
end

# Get the latest version for the given base version provided by #version
def hub_version
  base = version
  taginfo = JSON.parse(HTTParty.get("https://hub.docker.com/v2/repositories/#{username}/#{container_name}/tags/").body)['results']

  return { base: base, build: nil } if taginfo.nil?
  tags = []
  taginfo.each do |tag|
    tags << tag['name']
  end
  current_base = tags.grep(/#{base}/)
  return { base: base, build: nil } if current_base.empty?
  build = current_base.sort { |x, y|
    a = x.split('.')[base.split('.').count].to_i
    b = y.split('.')[base.split('.').count].to_i
    a <=> b
  }.last.split('.').last.to_i
  { base: base, build: build }
end

# return current hub version for the current base
def latest_hub_version
  latest = hub_version
  "#{latest[:base]}.#{latest[:build]}"
end

# return the next version for the current base
def next_version
  latest = hub_version
  base = version
  build = latest[:build] || -1
  build += 1
  "#{base}.#{build}"
end

task :install_deps do
  sh 'gem install bundler'
  sh 'bundle install'
end

desc 'login into Docker Hub'
task :login do
  sh "docker login -u #{ENV['DOCKER_USER']} -p #{ENV['DOCKER_PASS']}"
end

desc 'tags latest as next_version'
task :tag => :login do
  sh "docker tag #{username}/#{container_name}:latest #{username}/#{container_name}:#{next_version}"
end

desc 'pushes the next_version and latest to docker hub'
task :push => :tag do
  sh "docker push #{username}/#{container_name}:#{next_version}"
  sh "docker push #{username}/#{container_name}:latest"
end

desc 'builds as latest'
task :build => :install_deps do
  sh "docker build --rm=false -t #{username}/#{container_name}:latest ."
end

task default: [:build, :push]

But in the end both solutions does not provide me a Docker image per release channel and are limited in their own way. It looks hacky to me. I bet there are better approaches or tools I could use. I don't mind switching to something else completely.

What would be the best approach to accomplish something like what CoreOS does?

This Ansible role has been created and this .travis.yml has been used as an example in order to test the role. All services were able to start, except the sensu-client.

Attempts to solve the issue

In order to solve the issue a debug step was added to travis. When the build completed the following error was thrown:

The command "sudo docker exec "$(cat ${container_id})" ansible-playbook /etc/ansible/roles/role_under_test/tests/test-${SITE}.yml | grep -q 'changed=0.*failed=0' && (echo 'Idempotence test: pass' && exit 0) || (echo 'Idempotence test: fail' && exit 1)

" exited with 1.

0.07s$ sudo docker exec --tty "$(cat ${container_id})" env TERM=xterm sudo systemctl status sensu-client.service -l

● sensu-client.service - LSB: Sensu monitoring framework client

   Loaded: loaded (/etc/rc.d/init.d/sensu-client)

   Active: failed (Result: exit-code) since Mon 2016-08-08 00:18:20 UTC; 131ms ago

     Docs: man:systemd-sysv-generator(8)

  Process: 1534 ExecStart=/etc/rc.d/init.d/sensu-client start (code=exited, status=1/FAILURE)

Aug 08 00:18:19 c6354976113c systemd[1]: Starting LSB: Sensu monitoring framework client...

Aug 08 00:18:19 c6354976113c sensu-client[1534]: /etc/init.d/sensu-service: line 71: /etc/init.d/functions: No such file or directory

Aug 08 00:18:19 c6354976113c sensu-client[1534]: Starting sensu-client/etc/init.d/sensu-service: line 79: daemon: command not found

Aug 08 00:18:20 c6354976113c sensu-client[1534]: /etc/init.d/sensu-service: line 88: echo_failure: command not found

Aug 08 00:18:20 c6354976113c systemd[1]: sensu-client.service: control process exited, code=exited status=1

Aug 08 00:18:20 c6354976113c systemd[1]: Failed to start LSB: Sensu monitoring framework client.

Aug 08 00:18:20 c6354976113c systemd[1]: Unit sensu-client.service entered failed state.

Aug 08 00:18:20 c6354976113c systemd[1]: sensu-client.service failed.

When the ansible role is deployed on a remote system the sensu-client starts, but it fails on Travis although the other services are able to start.

According to this documentation it should be possible to add a travis badge to an ansible role in galaxy:

Push the changes to your GitHub repo. This will kick-off a Travis build. When the build completes, Travis will notify Galaxy, and your role will be automatically imported into Galaxy with a Travis badge

Although the steps were followed:

1. .travis.yml added to repo that contains the notification part
2. access token from travis profile added to ansible galaxy
3. role imported manually as it was not pushed automatically

the badge does not appear.

Discussion

The build itself does not show the notification. Other roles that have a travis badge do not show this notification either.

Questions

1. What is causing the issue?
2. Why does travis not say anything regarding the notification part?
3. How to debug it?