Skip to content

antonioua/ansible-gitlab-updater

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GitLab Omnibus in docker documentation

This document describes how to operate with gitlab in docker container

Motivation

This project was created to automated the update process of gtilab, to keep it up to date, simplify the restore procedure and moving all the infrastructure services to docker.

Contents

Installation

Prerequisites

Minimum OS version Centos 7
Our current setup has vms as described in table below:

FQDN IP
gitlab.mydomain.com 1.1.1.1
gitlab-ci-runner01.mydomain.com 1.1.1.2
gitlab-ci-runner02.mydomain.com 1.1.1.3
gitlab-ci-runner03.mydomain.com 1.1.1.4
gitlab-ci-runner04.mydomain.com 1.1.1.5
gitlab-ci-cache.mydomain.com 1.1.1.6

Docker host vm is accessible via port 2222 and it uses local pam authentication:

$ ssh gitlab.mydomain.com -p 2222

Docker installation

  1. Follow the instructions to uninstall all previous docker installs: https://docs.docker.com/install/linux/docker-ee/centos/#uninstall-old-versions
  2. Follow the instructions to install docker CE: https://docs.docker.com/install/linux/docker-ce/centos/#install-docker-ce
  3. Configure docker engine to allow insecure connection to registry, set proxy variables
$ mkdir -p /etc/systemd/system/docker.service.d/
$ vi /etc/systemd/system/docker.service.d/docker.conf

[Service]
ExecStart="--insecure-registry gitlab.mydomain.com:4567"
Environment="HTTP_PROXY=http://11129.50.100:8080" "HTTPS_PROXY=http://11129.50.100:8080" "NO_PROXY=*.mydomain.com"

$ systemctl daemon-reload
$ service docker restart

Run Gitlab in docker

On you first run you should start a container manually:

export gitlab_version="gitlab-10.6.2-ce.0"

docker run -d \
	--name ${gitlab_version} \
	--hostname gitlab.mydomain.com \
	--publish 443:443 \
	--publish 80:80 \
	--publish 22:22 \
	--publish 4567:4567 \
	--restart always \
	--volume /storage/docker_containers_data/${gitlab_version}/config:/etc/gitlab \
	--volume /storage/docker_containers_data/${gitlab_version}/logs:/var/log/gitlab \
	--volume /storage/docker_containers_data/${gitlab_version}/data:/var/opt/gitlab \
    --volume /storage/git-data:/mnt/git-data \
	--volume /git_backups:/var/opt/gitlab/backups \
	--volume /etc/localtime:/etc/localtime \
	--volume /storage/gitlab-registry:/var/opt/gitlab/gitlab-rails/shared/registry \
	gitlab/gitlab-ce:$(echo $gitlab_version | sed -e 's/gitlab-//')


$ docker ps

CONTAINER ID        IMAGE                          COMMAND             CREATED             STATUS                PORTS                                                                                  NAMES
45bdd74ebbb5        gitlab/gitlab-ce:10.6.2-ce.0   "/assets/wrapper"   4 days ago          Up 4 days (healthy)   0.0.0.0:22->22/tcp, 0.0.0.0:80->80/tcp, 0.0.0.0:443->443/tcp, 0.0.0.0:4567->4567/tcp   gitlab-10.6.2-ce.0

Please note!
You can stop and remove current container and start a new one in any time and do not loose any data.
All the data a stored on mounted volumes.

Gitlab CI runner installation

  1. Install the docker on vm for runner [Docker installation](#Docker installation)
  2. Install the runner:
$ ssh gitlab-ci-runner01.mydomain.com
$ yum install gitlab-ci-multi-runner
  1. Configure the runner:
$ vi /etc/gitlab-runner/config.toml

concurrent = 4
check_interval = 0

[[runners]]
  name = "gitlab-ci-runner01.mydomain.com"
  url = "https://gitlab.mydomain.com/"
  token = "880788aacd70f4e62588c9b37fc7ee"
  executor = "docker"
  shell = "bash"
  environment = ["http_proxy=http://11129.50.100:8080", "https_proxy=http://11129.50.100:8080", "no_proxy='127.0.0.1, localhost, *.mydomain.com'", "GIT_SSL_NO_VERIFY=true", "ad_user_for_external_api_calls=ta_jenkins_portal", "ad_password_for_external_api_calls=kXMkYYd11"]
  [runners.docker]
    tls_verify = false
    image = "gitlab.mydomain.com:4567/po/ci-tests:latest"
    privileged = true
    disable_cache = false
    volumes = ["/var/run/docker.sock:/var/run/docker.sock", "/cache", "/opt/docker_cache/ivy2/cache:/root/.ivy2/cache", "/opt/docker_cache/m2:/root/.m2"]
    shm_size = 0
  [runners.cache]
    Type = "s3"
    ServerAddress = "gitlab-ci-cache.mydomain.com:8000"
    AccessKey = "accessKey1"
    SecretKey = "verySecretKey1"
    BucketName = "runner"
    BucketLocation = ""
    Insecure = true
    Path = "gitlab/ci/cache"
    # To share the cache between two or more Runners, set the Shared flag to true.
    Shared = true

$ systemctl restart gitlab-runner.service
$ systemctl status gitlab-runner.service

ad_user_for_external_api_calls - variable is used to automatically upload artifcats to confluence in some jobs.

  1. Create dirs for cache volumes
mkdir -p /cache \
mkdir -p /opt/docker_cache/ivy2/cache \
mkdir -p /opt/docker_cache/m2
  1. Add gtilab certificates key chain to allow runner auth to gitlab registry.
    You should use cert from https://gitlab.mydomain.com/po/ansible-gitlab-provisioning/blob/master/roles/post_configuring/files/ssl/certificate_bundle.cer
$ ls -1 /etc/gitlab-runner/certs/
gitlab.mydomain.com.crt

Also I suggest to add the chanin to os keystore

$ vi /etc/pki/ca-trust/source/anchors/gitlab.mydomain.com.crt
$ update-ca-trust
  1. Register the new runner

Take a registration code from gtilab UI
On vm with runner run command below:

$ gitlab-runner register

choose:
https://gitlab.mydomain.com/
docker
gitlab.mydomain.com:4567/po/ci-tests:latest
  1. Add script for rotating old containers
$ vi /opt/scripts/clear-docker-cache.sh

#!/usr/bin/env bash


set -e

gitlab-ci-multi-runner stop >/dev/null 2>/dev/null

docker version >/dev/null 2>/dev/null

echo Clearing docker cache...

CONTAINERS=$(docker ps -a -q \
             --filter=status=exited \
             --filter=status=dead \
             --filter=label=com.gitlab.gitlab-runner.type=cache)

if [ -n "${CONTAINERS}" ]; then
    docker rm -v ${CONTAINERS}
fi

gitlab-ci-multi-runner start >/dev/null 2>/dev/null
  1. Add cron task
$ sudo crontab -e
0 23 * * 0   /opt/scripts/clear-docker-cache.sh

[ToDo] - Add task to install and configure runner with ansible

Amazon S3-compatible object storage server installation

Project repo: https://github.com/scality/S3

  1. Install and configure docker on gitlab-ci-cache.mydomain.com Docker installation
  2. Start a container with s3 service:
$ docker run -d --name s3-for-gitlabci -p 8000:8000 \
	-e ENDPOINT=gitlab-ci-cache.mydomain.com \
	-e SCALITY_ACCESS_KEY_ID=accessKey1 \
	-e SCALITY_SECRET_ACCESS_KEY=verySecretKey1 \
	-e S3BACKEND=mem \
	-e LOG_LEVEL=trace \
	--restart=always \
	scality/s3server:mem-latest

To check it just use

Gitlab Configuration

Config files and directories

Gitlab data files

/storage/
├── docker_containers_data
├── git-data
└── gitlab-registry

Dir with containers data

/storage/docker_containers_data/gitlab-10.6.2-ce.0/
├── config
├── data
└── logs

Dir for backups

$ ls -1 /git_backups/

1522621903_2018_04_02_10.6.2_gitlab_backup.tar
1522708338_2018_04_03_10.6.2_gitlab_backup.tar
1522794737_2018_04_04_10.6.2_gitlab_backup.tar
1522881087_2018_04_05_10.6.2_gitlab_backup.tar
1522967467_2018_04_06_10.6.2_gitlab_backup.tar
gitlab_backup_CONFIGS-2018_04_01.tar.gz
gitlab_backup_CONFIGS-2018_04_02.tar.gz
gitlab_backup_CONFIGS-2018_04_03.tar.gz
gitlab_backup_CONFIGS-2018_04_04.tar.gz
gitlab_backup_CONFIGS-2018_04_05.tar.gz
gitlab_backup_CONFIGS-2018_04_06.tar.gz

Gitlab data dir is specified in gitlab.rb

git_data_dirs({ "default" => { "path" => "/mnt/git-data" } })

Gitlab HTTPS support configuration.
Nginx from omnibus gitlab container is offloading https traffic

$ vi gitlab.rb

external_url 'https://gitlab.mydomain.com'
nginx['redirect_http_to_https'] = true

All the required dirs a mounted during container start as volumes

### Start new container
- name: Starting new container
  docker_container:
    name: "{{ next_img }}"
    image: "gitlab/gitlab-ce:{{ gitlab_img }}"
    hostname: "{{ hosti }}"
    restart_policy: always
    ports:
      - "22:22"
      - "80:80"
      - "443:443"
      - "4567:4567"
    volumes:
      - "/storage/docker_containers_data/{{ next_img }}/config:/etc/gitlab"
      - "/storage/docker_containers_data/{{ next_img }}/logs:/var/log/gitlab"
      - "/storage/docker_containers_data/{{ next_img }}/data:/var/opt/gitlab"
      - /storage/git-data:/mnt/git-data
      - /git_backups:/var/opt/gitlab/backups
      - /etc/localtime:/etc/localtime
      - /storage/gitlab-registry:/var/opt/gitlab/gitlab-rails/shared/registry
  when: mirroring|success

Registry configuration

Gitlab container registry configuration is shown below:

$ vi gitlab.rb

registry_external_url 'https://gitlab.mydomain.com:4567' <br />
gitlab_rails['gitlab_default_projects_features_container_registry'] = false <br />
registry_nginx['ssl_certificate'] = "/etc/gitlab/ssl/gitlab.mydomain.com.crt" <br />
registry_nginx['ssl_certificate_key'] = "/etc/gitlab/ssl/gitlab.mydomain.com.key" <br />

Backuping and rotating workflow

We have two types of backups:

  • full snapshots of gitlab vm --> are stored for 2 days
  • our own backups of gitlab application on separate mounted drive --> are stored for 5 days
$ crontab -l
5 1 * * * docker exec -i gitlab-10.6.2-ce.0 /bin/bash -c "/bin/tar czf /var/opt/gitlab/backups/gitlab_backup_CONFIGS-`date "+\%Y_\%m_\%d"`.tar.gz /etc/gitlab && /usr/bin/find /var/opt/gitlab/backups/ -type f -iname "gitlab_backup_CONFIGS-*" -mtime +5 | xargs rm -f"
#Ansible: Gitlab rake backup
10 1 * * * docker exec -t gitlab-10.6.2-ce.0 gitlab-rake gitlab:backup:create SKIP=artifacts,builds,uploads,registry

Our own backups are stored on mounted into docker volume, on host machine it is a separate mounted disk:

$ ls -1 /git_backups/
1522364301_2018_03_30_10.1.0_gitlab_backup.tar
1522434879_2018_03_30_10.1.0_gitlab_backup.tar
gitlab_backup_CONFIGS-2018_03_29.tar.gz
gitlab_backup_CONFIGS-2018_03_31.tar.gz
gitlab_backup_CONFIGS-2018_04_01.tar.gz

? What about configure raid for /git_backups/ https://erikugel.wordpress.com/2010/04/11/setting-up-linux-with-raid-faster-slackware-with-mdadm-and-xfs/

Rotating of backups is regulated in gitlab.rb config file by the following variable in seconds:

gitlab_rails['backup_keep_time'] = 432000

Backup of configs is made simply by creating an archive and rotating based on mtime value:

5 1 * * * docker exec -i gitlab-10.6.2-ce.0 /bin/bash -c "/bin/tar czf /var/opt/gitlab/backups/gitlab_backup_CONFIGS-`date "+\%Y_\%m_\%d"`.tar.gz /etc/gitlab && /usr/bin/find /var/opt/gitlab/backups/ -type f -iname "gitlab_backup_CONFIGS-*" -mtime +5 | xargs rm -f"

Provisioning of the cron jobs is implemented by ansible automatically and there is no need to do it manually after update.

Gitlab pages

[ToDo]

Maintenance and support

Docker container operations

  • Start stop, tail logs within a container
$ sudo docker exec -ti CONTAINER_NAME bash
root@gitlab:/# gitlab-ctl stop
root@gitlab:/# gitlab-ctl start
root@gitlab:/# gitlab-ctl tail
Ctr+D
  • Tail logs of container from host vm
$ sudo docker logs -f CONTAINER_NAME
  • Reconfigure gitlab in container (required if you have changed configuration, e.g. gitlab.rb)
    And you have two options here.

  • Reconfigure from host machine:

$ sudo docker restart gitlab
  • Reconfigure withing a container:
$ sudo docker exec -ti CONTAINER_NAME bash
root@gitlab:/# sudo gitlab-ctl reconfigure
  • Connect to postgress databae
$ docker exec -ti CONTAINER_NAME bash
$ su - gitlab-psql
$ psql -h /var/opt/gitlab/postgresql -d gitlabhq

Docker cleanups

  1. Remove not running containers
$ docker ps -a --filter "status=created" | awk '{ print $1 }' |xargs docker rm
  1. Remove unused images that is not in use by any container anymore
$ docker image prune -a
  1. Remove containers data dirs from host machine
$ ls -1 /storage/docker_containers_data/
gitlab-10.1.0-ce.0
gitlab-10.1.1-ce.0
gitlab-10.1.2-ce.0
gitlab-10.1.3-ce.0
gitlab-10.1.4-ce.0
gitlab-10.1.5-ce.0
gitlab-10.1.6-ce.0
gitlab-10.1.7-ce.0
gitlab-10.2.0-ce.0
gitlab-10.2.1-ce.0
gitlab-10.2.2-ce.0
gitlab-10.2.3-ce.0
gitlab-10.2.4-ce.0
gitlab-10.2.5-ce.0
gitlab-10.2.6-ce.0
gitlab-10.2.7-ce.0

Cleanup old blobs from gitalb registry

# docker exec -it CONTAINER_NAME /opt/gitlab/embedded/bin/registry garbage-collect /var/opt/gitlab/registry/config.yml

Or withing a container:

root@gitlab:/# gitlab-ctl registry-garbage-collect

There exist some solution for cleanup old blobs but it was not tested yet:
maybe this one works and should be tested https://gitlab.com/gitlab-org/docker-distribution-pruner

Gitlab CI Runner support

Cleanup old containers with cache volumes

We have a cron job installed on all runner vms earlier, so there is no need to cleanup it manually.
But in case you need to clean the docker follow the steps below.

Pause the runner via Gitlab UI and then follow the steps below

$ ssh gitlab-ci-runner01.mydomain.com
$ systemctl stop gitlab-runner.service

First remove all containers
$ docker rm `docker ps -aq`

Then remove orphaned volumes
$ docker volume rm $(docker volume ls -qf dangling=true)
$ systemctl start gitlab-runner.service

Gitlab container registry for FE tests

All the images used by Core and projects fro CI are stored in https://gitlab.mydomain.com/po/ci-tests/container_registry
How to push or modify existing images you can find thre: https://gitlab.mydomain.com/po/ci-tests/tree/latest

Update and restore Gitlab

Update Gitlab

This process is now automated.
You have two options:

  • run the commands manually
  • run jenkins job

Run the commands manually

If you are going to run the script manually then you should follow the steps:

Pull the freshest code with job http://jenkins-ci.mydomain.com/jenkins/view/service-jobs/job/ansible_gitlab_provisioning/

$ ssh to jenkins-ci.mydomain.com
$ cd /jobs/ansible_gitlab_provisioning/workspace

Get all available versions
$ export http_proxy=http://11129.50.100:8080 https_proxy=http://11129.50.100:8080
$ curl -s https://registry.hub.docker.com/v1/repositories/gitlab/gitlab-ce/tags | python -mjson.tool | grep name | sort

Modify file with list of versions to update
$ vi update.list

Run the command
$ while read gitlab_img; do \
	echo $gitlab_img; \
	ansible-playbook -i hosts/hosts.ini gitlab_updating.yml -e "hosti=gitlab.mydomain.com gitlab_img=$gitlab_img"; \
	sleep 60; \
done < update.list

Run jenkins job

Please NOTE!!! Be careful and follow jenkins job console logs for versions update<br/ > To run jenkins job go to http://jenkins-ci.mydomain.com/jenkins/view/service-jobs/job/ansible_gitlab_provisioning/ select versions to update and press build button.

Restore Gitlab

The official documentain is there: https://docs.gitlab.com/ee/raketasks/backup_restore.html
Please note that we are using Omnibus installation

Here we have two choices:

  • Restore to certain version of the docker container (most usable)
  • Full restore with all repository data

Restore to certain version of the docker container

This will return us to certain state of postgres db and gitlab configuration

  1. Make sure the container is stopped
  2. Start the required container with command described there Run Gitlab in docker
  3. Check logs if there no errors:
$ sudo docker logs -f CONTAINER_NAME

Full restore with all repository data

  1. Make sure the container is stopped
  2. Start the container of certain version. The container version should match your backup version. Run Gitlab in docker
  3. Stop gitlab inside a container:
root@gitlab:/# sudo gitlab-ctl stop unicorn
root@gitlab:/# sudo gitlab-ctl stop sidekiq

Verify
root@gitlab:/# sudo gitlab-ctl status
  1. Make sure the required backup archive is visible within a container:
$ docker exec -ti CONTAINER_NAME bash -c "ls -1 /var/opt/gitlab/backups/"
1522621903_2018_04_02_10.6.2_gitlab_backup.tar
  1. Restore all gitlab data from archive with rake:
# This command will overwrite the contents of your GitLab database!
$ docker exec -ti CONTAINER_NAME bash -c "gitlab-rake gitlab:backup:restore BACKUP=1522621903_2018_04_02_10.6.2"
  1. Next, restore /etc/gitlab/gitlab-secrets.json if necessary.

  2. Restart and check GitLab:

$ sudo docker exec -ti CONTAINER_NAME bash
root@gitlab:/# gitlab-ctl restart
root@gitlab:/# sudo gitlab-rake gitlab:check SANITIZE=true

If there is a GitLab version mismatch between your backup tar file and the installed version of GitLab, the restore command will abort with an error.
Run the correct GitLab container version and try again Run Gitlab in docker

Troubleshooting

  • Error produced by ansible: Either path or file obj needs to be provided.

Resource: https://www.reddit.com/r/ansible/comments/7cq9q5/issue_with_docker_container_error_pulling_image/

Fix: on your ansible server install the docker-python.x86_64

  • If your docker do not trust CA then add it:
$ vi /etc/docker/certs.d/pt_cas.crt
$ systemctl restart docker

[ToDo]

Usefull stuff

If you need to run certain task
$ ansible-playbook playbook.yml --start-at-task="install packages" <br />

If you need to run playbook with confirmation of each step/task:
$ ansible-playbook playbook.yml --step <br />

What to do on nearest update

  1. Enable in gtilab.rb
##! **Recommended by: https://raymii.org/s/tutorials/Strong_SSL_Security_On_nginx.html
##!                   https://cipherli.st/**
nginx['ssl_protocols'] = "TLSv1 TLSv1.1 TLSv1.2"
  1. Fix https issue - Fixed

2.1 Rollback Anton Nazin's change in nginx conf i docker:

root@gitlab:/# vi /var/opt/gitlab/nginx/conf/gitlab-http.conf +83

#  add_header Strict-Transport-Security "max-age=31536000";

2.1 We need a DNS zone for *.gitlab.mydomain.com, to be able to serve on the same ip requests with HOST header like: gitlab.mydomain.com,, vitalii.rakovtsii.gitlab.mydomain.com

2.2 We need a single TLS certificate for *.gitlab.mydomain.com and gitlab.mydomain.com with Alternative Name record inside (certificate+private key), Multidomain/wildcard certificate.

X509v3 Subject Alternative Name: DNS:*.gitlab.mydomain.com, DNS:gitlab.mydomain.com

This is needed to resolve chrome's issue: "The certificate for this site does not contain a Subject Alternative Name extension containing a domain name or IP address."

https://www.digicert.com/subject-alternative-name.htm

SD ticket: IN0801458

  1. reconfigure docker too use devicemapper driver or overlay, read the suggestions for centos 7.4
$ vi /etc/docker/daemon.json
{
  "storage-driver": "devicemapper",
  "storage-opts": [
    "dm.directlvm_device=???",
    "dm.thinp_percent=95",
    "dm.thinp_metapercent=1",
    "dm.thinp_autoextend_threshold=80",
    "dm.thinp_autoextend_percent=20",
    "dm.directlvm_device_force=false"
  ]
}

Or:

{
    "data-root": "/mnt/docker_data",
    "storage-driver": "overlay"
}

systemctl daemon-reload
systemctl restart docker
  1. Mount a dir for gitlab pages as a volume:
  2. Add to Readme about Gitlab pages configuration in .rb file
  3. How to test s3 from client: use aws cli agent, Please note that to be able to store cahce you need to create a bucket on s3 manualy!

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published