UNCLASSIFIED

You need to sign in or sign up before continuing.
Commits (10)
...@@ -17,6 +17,13 @@ If you need to contact the Container Hardening team, please identify your assign ...@@ -17,6 +17,13 @@ If you need to contact the Container Hardening team, please identify your assign
If you have no assignee, feel free to tag Container Hardening leadership in your issue by commenting on this issue with your questions/concerns and then add `/cc @ironbank-notifications/leadership`. Gitlab will automatically notify all Container Hardening leadership to look at this issue and respond. If you have no assignee, feel free to tag Container Hardening leadership in your issue by commenting on this issue with your questions/concerns and then add `/cc @ironbank-notifications/leadership`. Gitlab will automatically notify all Container Hardening leadership to look at this issue and respond.
## Get Unstuck/AMA:
Iron Bank Get Unstuck/AMA Working Sessions every Wednesday from 1630-1730EST.
Need some help with your containers getting through Iron Bank? Have questions on where things are at? Are you feeling stuck and want to figure out the next steps? This is the meeting for you! Come meet with the Iron Bank leadership and engineers to get answers to your questions.
Register in advance for this meeting: https://www.zoomgov.com/meeting/register/vJIsf-ytpz8qHSN_JW8Hl9Qf0AZZXSCSmfo
After registering, you will receive a confirmation email containing information about joining the meeting.
If you have any questions, please come to our Get Unstuck/AMA sessions. There we will have the right combination of business folks and engineers to get your questions answered.
## Responsibilities ## Responsibilities
...@@ -25,28 +32,140 @@ If this application is owned by a Contributor or Vendor (identifed as `Owner::Co ...@@ -25,28 +32,140 @@ If this application is owned by a Contributor or Vendor (identifed as `Owner::Co
## Definition of Done ## Definition of Done
Hardening:
- [ ] Hardening manifest is created and adheres to the schema (https://repo1.dsop.io/ironbank-tools/ironbank-pipeline/-/blob/master/schema/hardening_manifest.schema.json)
- [ ] Container builds successfully through the Gitlab CI pipeline
- [ ] Branch has been merged into `development`
- [ ] Project is configured for automatic renovate updates (if possible)
Justifications: This checklist is meant to provide a high level overview of the process and steps for getting your container(s) onto Iron Bank.
- [ ] All findings have been justified per the above documentation
- [ ] Create a Repo1 account (https://repo1.dso.mil/users/sign_in) to get access to the public repository of containers. You can register by clicking on the 'Sign in with Iron Bank SSO' button in the sign-in page, followed by the Register button
- [ ] Fill out the onboarding form: https://p1.dso.mil/#/products/iron-bank/getting-started
- [ ] Attend our once weekly onboarding session where you can ask questions. [Register here](https://www.zoomgov.com/meeting/register/vJIsce6rpzkqGq9hHHRscNfGENYqvRL1s10%E2%81%A9).
- [ ] Your Onboarding form will be processed by the Iron Bank team, who will then assign it a priority level and create your repository. You will receive an email that your Gitlab issue has been created and is ready for you to complete the hardening process
- [ ] Ensure that all POCs are assigned to the issue to ensure proper tracking and notifications
## Hardening Process
### Repository Requirements
[Full documentation](https://repo1.dso.mil/dsop/dccscr/-/blob/master/Hardening/structure_requirements.md)
- [ ] A Dockerfile has been created in the root of the repository
- [ ] Hardening_manifest.yaml has been created in the root of the repository
- [ ] The project has a LICENSE or a copy of the EULA
- [ ] The project has a README in the root of the repository with sufficient instructions on using the Iron Bank version of the image
- [ ] If your container is an enterprise/commercial container, the opensource version is ready
- [ ] Scripts used in the Dockerfile are placed into a `scripts` directory
- [ ] Configuration files are placed into a `config` directory
- [ ] Project is [configured for automatic renovate updates](https://repo1.dso.mil/dsop/dccscr/-/blob/master/Hardening/Renovate.md) (if possible)
- [ ] Renovate.json is present in root of repository
- [ ] Reviewers have been specified for notifications on new merge requests
### Dockerfile Requirements
[Full documentation](https://repo1.dso.mil/dsop/dccscr/-/blob/master/Hardening/Dockerfile_Requirements.md)
- [ ] There is one Dockerfile named Dockerfile
- [ ] The Dockerfile has the BASE_REGISTRY, BASE_IMAGE, and BASE_TAG arguments (used for local builds; the values in hardening_manifest.yaml are what will be used in the Container Hardening Pipeline)
- [ ] The Dockerfile is [based on a hardened Iron Bank image](https://repo1.dso.mil/dsop/dccscr/-/blob/master/Hardening/Dockerfile_Requirements.md#requirements)
- [ ] The Dockerfile includes a HEALTHCHECK (required if it is an application container)
- [ ] The Dockerfile starts the container as a non-root USER. Otherwise, if you must run as root, you must have proper justification.
- [ ] If your ENTRYPOINT entails using a script, the script is copied from a scripts directory on the project root
- [ ] No ADD instructions are used in the Dockerfile
## Hardening Manifest
[Full documentation](https://repo1.dso.mil/dsop/dccscr/-/tree/master/hardening%20manifest)
- [ ] Begin with this example and update with relevant information: https://repo1.dso.mil/dsop/dccscr/-/blob/master/hardening%20manifest/hardening_manifest.yaml
- [ ] Hardening manifest adheres to the following schema: https://repo1.dsop.io/ironbank-tools/ironbank-pipeline/-/blob/master/schema/hardening_manifest.schema.json
- [ ] The BASE_IMAGE and BASE_TAG arguments refer to a hardened/approved Iron Bank image (BASE_REGISTRY defaults to `registry1.dso.mil/ironbank` in the pipeline)
- [ ] Relevant image metadata has been entered for the corresponding labels
- [ ] Any downloaded resources include a checksum for verification (letters must be lowercase)
- [ ] For resource URLs that require authentication, credentials have been provided to an Iron Bank team member
- [ ] The maintainers' contact information has been provided in the `maintainers` section
## Gitlab CI Pipeline
[Full documentation](https://repo1.dso.mil/dsop/dccscr/-/tree/master/pipeline)
- [ ] Validate your container builds successfully through the Gitlab CI pipeline. When viewing the repository in repo1.dso.mil, go to `CI/CD > Pipelines` on the left. From there, you can see the status of your pipelines.
- [ ] Review scan output from `csv output` stage of the pipeline. For instructions on downloading the findings spreadsheet, click [here](https://repo1.dso.mil/dsop/dccscr/-/blob/master/pre-approval/spreadsheet.md)
- [ ] Fix vulnerabilities that were found and run the pipeline again before requesting a merge to the development branch
## Pre-Approval:
[Full documentation](https://repo1.dso.mil/dsop/dccscr/-/tree/master/pre-approval)
- [ ] Submit a Merge Request to the development branch
- [ ] Feature branch has been merged into development
- [ ] All findings from the development branch pipeline have been justified per the above documentation
- [ ] Justifications have been attached to this issue - [ ] Justifications have been attached to this issue
- [ ] Apply the label `Approval` to indicate this container is ready for the approval phase
Note: The justifications must be provided in a timely fashion. Failure to do so could result in new findings being identified which may start this process over. - [ ] Apply the `Approval` label and remove the `Doing` label to indicate this container is ready for the approval phase
_Note: The justifications must be provided in a timely fashion. Failure to do so could result in new findings being identified which may start this process over._
## Approval Process (Container Hardening Team processes):
[Full documentation](https://repo1.dso.mil/dsop/dccscr/-/tree/master/approval)
Approval Process (Container Hardening Team processes):
- [ ] Peer review from Container Hardening Team - [ ] Peer review from Container Hardening Team
- [ ] Findings Approver has reviewed and approved all justifications - [ ] Findings Approver has reviewed and approved all justifications
- [ ] Approval request has been sent to Authorizing Official - [ ] Approval request has been sent to Authorizing Official
- [ ] Approval request has been processed by Authorizing Official - [ ] Approval request has been processed by Authorizing Official
Note: If the above approval process is kicked back for any reason, the `Approval` label will be removed and the issue will be sent back to `Open`. Any comments will be listed in this issue for you to address. Once they have been addressed, you may re-add the `Approval` label. One of the following statuses is assigned:
- [ ] Conditional approval has been granted by the Authorizing Official for this container (`Approval::Expiring` label is applied)
- [ ] This container has been approved by the Authorizing Official (`Approved` label is applied)
_Note: If the above approval process is kicked back for any reason, the `Approval` label will be removed and the issue will be sent back to `Open`. Any comments will be listed in this issue for you to address. Once they have been addressed, you may re-add the `Approval` label._
## Post-Approval
[Full documentation](https://repo1.dso.mil/dsop/dccscr/-/tree/master/post%20approval)
- [ ] Your issue has been closed
- [ ] Your project has been merged into master
- [ ] Master branch pipeline has completed successfully (at this point, the image is made available on `ironbank.dso.mil` and `registry1.dso.mil` )
_Note: Now that your application has been approved, your container(s) will be subjected to continuous monitoring. If new CVEs are discovered or bugs are identified, you will need to address the issues and return to step 5 (Gitlab CI Pipeline). As you make changes, please make sure you are adhering to all of the requirements of the hardening process._
## Post Approval
### Continuous Monitoring ### Continuous Monitoring
......
ARG GITLAB_VERSION=v14.2.0-ubi8
ARG BASE_REGISTRY=registry1.dso.mil
ARG BASE_IMAGE=gitlab/gitlab/gitlab-rails
ARG BASE_TAG=14.2.0
ARG RAILS_IMAGE=${BASE_REGISTRY}/${BASE_IMAGE}:${BASE_TAG}
FROM ${RAILS_IMAGE}
ARG GITLAB_VERSION
ARG GITLAB_USER=git
ARG DNF_OPTS
ENV LIBDIR ${LIBDIR:-"/usr/lib64"}
ADD gitlab-toolbox-ee.tar.gz /
ADD gitlab-python.tar.gz /
COPY scripts/bin/* /usr/local/bin/
COPY scripts/lib/* ${LIBDIR}/ruby/vendor_ruby/
RUN dnf clean all \
&& rm -r /var/cache/dnf \
&& dnf ${DNF_OPTS} install -by --nodocs ca-certificates openssl
USER ${GITLAB_USER}:${GITLAB_USER}
ENTRYPOINT ["/usr/local/bin/entrypoint.sh"]
MIT License
Copyright (c) 2017
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
# <application name> # gitlab-toolbox-container
Project template for all Iron Bank container repositories. Gitlab Toolbox is an entry point for interaction with other containers in the cluster. It contains scripts for running Rake tasks, backup, restore, and tools to intract with object storage.
\ No newline at end of file
## GitLab Build
The hardened containers for GitLab require the gitlab correctly version assets blob to be available for download.
There are some included scripts to make this easier for user building images offline on their machine.
## Building
1. Switch to the desired versioned directory
2. Run `./build-scripts/build.sh`
* Runs docker build, and takes care of setting the appropriate build args for users running locally
`build-script/build.sh` is provided as an example and helpful for building locally. You can also instead call docker build and pass build-args directly.
## Build Phases
Some of the GitLab containers are build ontop of previous containers, building the containers in ordered phases is necessary to build all containers.
- Phase One
* kubectl
* gitlab-ruby
* gitlab-container-registry
- Phase Two
* git-base
* gitlab-exporter
* gitlab-mailroom
* gitlab-shell
* gitlab-rails
* gitlab-workhorse
- Phase 3
* gitaly
- Phase 4
* gitlab-sidekiq
* gitlab-toolbox
* gitlab-webservice
#!/bin/bash
# NOTICE: This script requires `docker`.
set -euxo pipefail
TAG=${1:-14.2.0}
REPOSITORY=${2:-}
DOCKER_OPTS=${DOCKER_OPTS:-""}
imageName() {
printf "${REPOSITORY}${1}:${TAG}"
}
buildImage() {
IMAGE="${1}"
CONTEXT="${IMAGE%*-ee}"
{
docker build \
-t "$(imageName ${IMAGE})" . \
${DOCKER_OPTS:-} | tee ${CONTEXT}.out
} || {
echo "${CONTEXT}" >> failed.log
}
}
# Cleanup log outputs from previous build
rm -f *.out failed.log
DOCKER_OPTS="$DOCKER_OPTS --build-arg RAILS_IMAGE=$(imageName gitlab-rails)"
buildImage gitlab-toolbox
#!/bin/bash
set -euxo pipefail
rm -f *.tar.gz *.out failed.log
apiVersion: v1
# The repository name in registry1, excluding /ironbank/
name: "gitlab/gitlab/gitlab-toolbox"
# List of tags to push for the repository in registry1
# The most specific version should be the first tag and will be shown
# on ironbank.dsop.io
tags:
- "14.2.0"
- "latest"
# Build args passed to Dockerfile ARGs
args:
BASE_IMAGE: "gitlab/gitlab/gitlab-rails"
BASE_TAG: "14.2.0"
# Docker image labels
labels:
org.opencontainers.image.title: "Gitlab Toolbox"
## Human-readable description of the software packaged in the image
org.opencontainers.image.description: "GitLab Toolbox is an entry point for interaction with other containers in the cluster. It contains scripts for running Rake tasks, backup, restore, and tools to intract with object storage."
## License(s) under which contained software is distributed
org.opencontainers.image.licenses: "MIT License"
## URL to find more information on the image
org.opencontainers.image.url: "https://about.gitlab.com/"
## Name of the distributing entity, organization or individual
org.opencontainers.image.vendor: "Gitlab"
org.opencontainers.image.version: "14.2.0"
## Keywords to help with search (ex. "cicd,gitops,golang")
mil.dso.ironbank.image.keywords: "gitlab, git, gitops"
## This value can be "opensource" or "commercial"
mil.dso.ironbank.image.type: "commercial"
## Product the image belongs to for grouping multiple images
mil.dso.ironbank.product.name: "gitlab"
# List of project maintainers
# FIXME: Fill in the following details for the current container owner in the whitelist
# FIXME: Include any other vendor information if applicable
maintainers:
- email: "dj@gitlab.com "
# # The name of the current container owner
name: "DJ Mountney"
# # The gitlab username of the current container owner
username: "twk3"
# cht_member: true # FIXME: Uncomment if the maintainer is a member of CHT
- name: "Al Fontaine"
username: "alfontaine"
email: "alan.fontaine@centauricorp.com"
- email: "adam.martin@rancherfederal.com"
name: "Adam Martin"
username: "adam.martin"
cht_member: true
resources:
- url: "https://gitlab-ubi.s3.amazonaws.com/ubi8-build-dependencies-v14.2.0-ubi8/gitlab-toolbox-ee.tar.gz"
filename: "gitlab-toolbox-ee.tar.gz"
validation:
type: "sha256"
value: "d0b2020e9ba41367ea3bb36e1fabc924ca7fb8241224e386924f2822aa5a44a4"
- url: "https://gitlab-ubi.s3.amazonaws.com/ubi8-build-dependencies-v14.2.0-ubi8/gitlab-python.tar.gz"
filename: "gitlab-python.tar.gz"
validation:
type: "sha256"
value: "c7c279f90687febbfb537c17f43b6db9dc0d1916feb7e99bdd40c17e61fe0064"
io.openshift.tags=gitlab-toolbox
io.openshift.wants=gitlab-webservice
io.k8s.description=Container with tools to diagnose GitLab application
issues.
io.openshift.non-scalable=false
io.openshift-min-memory=350Mi
io.openshift.min-cpu=50m
#!/bin/bash
set -e
ACTION=""
export BACKUP_BUCKET_NAME=${BACKUP_BUCKET_NAME-gitlab-backups}
export BACKUP_BACKEND=${BACKUP_BACKEND-s3}
S3_CMD_BACKUP_OPTION=""
rails_dir=/srv/gitlab
backups_path=$rails_dir/tmp/backups
backup_tars_path=$rails_dir/tmp/backup_tars
object_storage_backends=( registry uploads artifacts lfs packages external_diffs terraform_state pages )
skipping_backup_for=()
function usage()
{
cat << HEREDOC
Usage: backup-utility [--restore|--cleanup] [-f URL] [-t TIMESTAMP] [--skip COMPONENT] [--backend BACKEND] [--s3config CONFIG]
Options:
-h, --help Show this help message and exit.
--restore [-t TIMESTAMP | -f URL] When specified, utility restores from an existing backup specified
as url or timestamp in object storage.
-f URL http(s):/ftp:/file: URL with backup location. Use with --restore.
-t TIMESTAMP Timestamp (part before '_gitlab_backup.tar' in archive name),
can be used to specify backup source or target name.
--rsyncable Pass the '--rsyncable' parameter to gzip for artifact compression.
--skip COMPONENT When specified, utility will skip the backup of COMPONENT.
May be defined multiple times. Valid values for COMPONENT are
db, repositories, and any of the object storages (e.g. 'lfs').
--backend BACKEND Object storage backend to use for backups.
Can be either 's3' or 'gcs'.
--s3config CONFIG S3 backend configuration to use for backups storage.
Special config file for s3cmd (see: https://s3tools.org/usage)
--storage-class CLASSNAME Pass this storage class to the gcs or s3cmd for more cost-efficient
storage of backups.
--maximum-backups N Only keep the most recent N number of backups, deleting others after success.
Requires s3config credentials to be able to list and delete objects.
--cleanup Run the backup cleanup without creating a new backup. Can be used with the
'maximum-backups' option to clean old remote backups.
HEREDOC
}
# Checks if provided argument is a url for downloading it
function is_url() {
regex='(https?|ftp|file)://[-A-Za-z0-9\+&@#/%?=~_|!:,.;]*[-A-Za-z0-9\+&@#/%=~_|]'
[[ $1 =~ $regex ]]
}
function fetch_remote_backup(){
mkdir -p $backups_path
output_path=$backups_path/0_gitlab_backup.tar
if is_url $1; then
>&2 echo "Downloading from $1";
curl --retry 6 --progress-bar -o $output_path $1
else # It's a timestamp
file_name="$1_gitlab_backup.tar"
if [ "${BACKUP_BACKEND}" = "s3" ]; then
s3cmd ${S3_CMD_BACKUP_OPTION} get "s3://$BACKUP_BUCKET_NAME/$file_name" $output_path > /dev/null
elif [ "${BACKUP_BACKEND}" = "gcs" ]; then
gsutil cp "gs://$BACKUP_BUCKET_NAME/$file_name" $output_path > /dev/null
else
echo "Unknown backend: ${BACKUP_BACKEND}"
fi
fi
echo $output_path
}
function unpack_backup(){
local file_path=$1
cd $(dirname $file_path)
echo "Unpacking backup"
if [ ! -f $file_path ]; then
echo $file_path not found
exit 1
fi
tar -xf $file_path
}
function pack_backup(){
echo "Packing up backup tar"
local backup_name=$1
tar -cf ${backup_tars_path}/${backup_name}.tar -C $backups_path .
}
function get_version(){
cat $rails_dir/VERSION
}
function get_backup_name(){
if [ -n "$BACKUP_TIMESTAMP" ]; then
echo ${BACKUP_TIMESTAMP}_gitlab_backup
else
now_timestamp=$(date +%s_%Y_%m_%d)
gitlab_version=$(get_version)
echo ${now_timestamp}_${gitlab_version}_gitlab_backup
fi
}
function get_existing_backups(){
# This will only match backups with the same naming convention as backups generated by this script
# Example: TIMESTAMP_YYYY_MM_DD_VERSION_gitlab_backup.tar
case $BACKUP_BACKEND in
s3)
existing_backups=($(s3cmd ${S3_CMD_BACKUP_OPTION} ls s3://$BACKUP_BUCKET_NAME --rinclude '^\d{10}_\d{4}_\d{2}_\d{2}_.+_gitlab_backup.tar$' | awk '{print $4}' | LC_ALL=C sort))
;;
gcs)
# Note: gsutil doesn't support regex, so we need to try to match the prefix as best we can with wildcards
# https://cloud.google.com/storage/docs/gsutil/addlhelp/WildcardNames#other-wildcard-characters
existing_backups=($(gsutil ls gs://$BACKUP_BUCKET_NAME/[0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9]_[0-9][0-9][0-9][0-9]_[0-9][0-9]_[0-9][0-9]_\*_gitlab_backup.tar | LC_ALL=C sort))
;;
*)
echo "Unknown backend for backup: ${BACKUP_BACKEND}"
exit 1
;;
esac
}
function remove_backup(){
local backup_to_remove=$1
if [ "${BACKUP_BACKEND}" = "s3" ]; then
s3cmd ${S3_CMD_BACKUP_OPTION} del ${backup_to_remove} > /dev/null
elif [ "${BACKUP_BACKEND}" = "gcs" ]; then
gsutil rm ${backup_to_remove} > /dev/null
else
echo "Unknown backend for backup: ${BACKUP_BACKEND}"
exit 1
fi
}
function cleanup(){
rm -rf $backups_path/*
rm -rf $backup_tars_path/*
if [ -n "$MAXIMUM_BACKUPS" ]; then
get_existing_backups
echo "Found ${#existing_backups[@]} existing backups. Maximum allowed is $MAXIMUM_BACKUPS"
if [ ${#existing_backups[@]} -gt $MAXIMUM_BACKUPS ]; then
i=0
while [ $i -lt $(expr ${#existing_backups[@]} - $MAXIMUM_BACKUPS) ]; do
echo "Deleting old backup ${existing_backups[$i]}"
remove_backup ${existing_backups[$i]}
((++i))
done
fi
echo "[DONE] Finished pruning old backups"
fi
}
function write_backup_info(){
cat << EOF > $backups_path/backup_information.yml
:db_version: $($rails_dir/bin/rails runner "File.write('/tmp/db_version', ActiveRecord::Migrator.current_version.to_s)" && cat /tmp/db_version)
:backup_created_at: $(date "+%Y-%m-%d %H:%M:%S %z")
:gitlab_version: $(get_version)
:tar_version: $(tar --version | head -n 1)
:installation_type: gitlab-helm-chart
:skipped: $1
EOF
}
function get_skipped(){
all=( builds.tar.gz db repositories pages.tar.gz )
for storage in ${object_storage_backends[@]}; do
all+=( "${storage}.tar.gz" );
done;
skipped_string=""
for backup_item in ${all[@]}; do
if [ ! -e $backups_path/$backup_item ]; then
skipped_string="$skipped_string,${backup_item%.tar.gz}";
fi;
done;
echo ${skipped_string#,}
}
function backup(){
backup_name=$(get_backup_name)
mkdir -p $backup_tars_path $backups_path
if ! [[ ${skipping_backup_for[@]} =~ "db" ]]; then
gitlab-rake gitlab:backup:db:create
fi
if ! [[ ${skipping_backup_for[@]} =~ "repositories" ]]; then
gitlab-rake gitlab:backup:repo:create
fi
for backup_item in ${object_storage_backends[@]}; do
if ! [[ ${skipping_backup_for[@]} =~ $backup_item ]]; then
object-storage-backup $backup_item $backups_path/${backup_item}.tar.gz
fi
done
skipped=$(get_skipped $backup_name)
write_backup_info $skipped
pack_backup $backup_name
if [ "${BACKUP_BACKEND}" = "s3" ]; then
if [ -z "${STORAGE_CLASS}" ]; then
s3cmd ${S3_CMD_BACKUP_OPTION} put ${backup_tars_path}/${backup_name}.tar s3://$BACKUP_BUCKET_NAME > /dev/null
else
s3cmd ${S3_CMD_BACKUP_OPTION} put --storage-class "${STORAGE_CLASS}" ${backup_tars_path}/${backup_name}.tar s3://$BACKUP_BUCKET_NAME > /dev/null
fi
echo "[DONE] Backup can be found at s3://$BACKUP_BUCKET_NAME/${backup_name}.tar"
elif [ "${BACKUP_BACKEND}" = "gcs" ]; then
if [ -z "${STORAGE_CLASS}" ]; then
gsutil cp -n ${backup_tars_path}/${backup_name}.tar gs://$BACKUP_BUCKET_NAME > /dev/null
else
gsutil cp -s "${STORAGE_CLASS}" -n ${backup_tars_path}/${backup_name}.tar gs://$BACKUP_BUCKET_NAME > /dev/null
fi
echo "[DONE] Backup can be found at gs://$BACKUP_BUCKET_NAME/${backup_name}.tar"
else
echo "Unknown backend for backup: ${BACKUP_BACKEND}"
fi
cleanup
}
function is_skipped() {
[[ $SKIPPED =~ $1 ]]
}
function restore(){
if [ -z "$BACKUP_URL" ] && [ -z "$BACKUP_TIMESTAMP" ]; then
echo "You need to set BACKUP_URL or BACKUP_TIMESTAMP variable"
exit 1
fi
BACKUP=${BACKUP_URL-}
if [ -z "$BACKUP" ]; then
BACKUP=$BACKUP_TIMESTAMP
fi
file=$(fetch_remote_backup $BACKUP)
dir_name=$(dirname $file)
file_name=$(basename $file)
timestamp="${file_name%%_*}"
export BACKUP=$timestamp
unpack_backup $file
skipped_line=$(grep skipped $(dirname $file)/backup_information.yml)
export SKIPPED=$(echo ${skipped_line#:skipped:})
installation_type_line=$(grep installation_type $(dirname $file)/backup_information.yml || echo ":installation_type: unknown")
export INSTALLATION_TYPE=$(echo ${installation_type_line#:installation_type: })
! is_skipped "db" && gitlab-rake gitlab:db:drop_tables
! is_skipped "db" && gitlab-rake gitlab:backup:db:restore
# Previous versions of the dump failed to mark the repos as skipped, so we additionally check for the directory
if [ -e $backups_path/repositories ]; then
! is_skipped "repositories" && gitlab-rake gitlab:backup:repo:restore
fi
! is_skipped "builds" && gitlab-rake gitlab:backup:builds:restore
if [ "$INSTALLATION_TYPE" = "gitlab-helm-chart" ]; then
for restore_item in ${object_storage_backends[@]}; do
if [ -f $backups_path/${restore_item}.tar.gz ]; then
! is_skipped $restore_item && object-storage-restore $restore_item $backups_path/${restore_item}.tar.gz
fi
done
else
echo "Backup tarball not from a Helm chart based installation. Not processing files in object storage."
fi
gitlab-rake cache:clear
}
while [[ $# -gt 0 ]]
do
key="$1"
case $key in
-h|--help)
usage
ACTION="none"
break
;;
-f|--file)
BACKUP_URL="$2"
shift
shift
;;
-t|--timestamp)
BACKUP_TIMESTAMP="$2"
shift
shift
;;
--backend)
export BACKUP_BACKEND="$2"
shift
shift
;;
--s3config)
if [ ! -f $2 ]; then
echo "s3cmd file specified does not exist";
exit 1;
fi
export S3_CMD_BACKUP_OPTION="--config=$2 "
shift
shift
;;
--restore)
if [ -z "$ACTION" ]; then
ACTION="restore"
else
echo "Only one action at a time is supported"
exit 1
fi
shift
;;
--rsyncable)
export GZIP_RSYNCABLE="yes"
shift
;;
--skip)
skipping_backup_for+=( "$2" )
shift
shift
;;
--storage-class)
export STORAGE_CLASS="$2"
shift
shift
;;
--maximum-backups)
export MAXIMUM_BACKUPS="$2"
if ! [[ $MAXIMUM_BACKUPS =~ ^-?[0-9]+$ ]]; then
echo "Value specified for --maximum-backups must be an integer. Got: ${MAXIMUM_BACKUPS}"
exit 1
fi
shift
shift
;;
--cleanup)
if [ -z "$ACTION" ]; then
ACTION="cleanup"
else
echo "Only one action at a time is supported"
exit 1
fi
shift
;;
*)
usage
echo "Unexpected parameter: $key"
exit 1
;;
esac
done
if [ "$ACTION" = "restore" ]; then
restore
elif [ "$ACTION" = "cleanup" ]; then
cleanup
elif [ -z "$ACTION" ]; then
ACTION="backup"
backup
fi
#!/bin/bash
set -e
/scripts/set-config "${CONFIG_TEMPLATE_DIRECTORY}" "${CONFIG_DIRECTORY:=$CONFIG_TEMPLATE_DIRECTORY}"
cd /srv/gitlab;
echo "Attempting to run '$@' as a main process";
exec "$@";
#!/bin/bash
rails_dir=/srv/gitlab
cd $rails_dir
$rails_dir/bin/bundle exec rails "$@"
#!/bin/bash
rails_dir=/srv/gitlab
cd $rails_dir
$rails_dir/bin/bundle exec rake -f $rails_dir/Rakefile "$@"
#!/usr/bin/env ruby
require 'object_storage_backup'
abort ("backup_item and output_tar_path arguments needs to be passed to the script") unless ARGV.length == 2
bucket_name = ENV["#{ARGV[0].upcase}_BUCKET_NAME"] || "gitlab-#{ARGV[0]}"
tmp_bucket = ENV['TMP_BUCKET_NAME'] || 'tmp'
backend_type = ENV['BACKUP_BACKEND'] || 's3'
ObjectStorageBackup.new(ARGV[0], ARGV[1], bucket_name, tmp_bucket, backend_type).backup
#!/usr/bin/env ruby
require 'object_storage_backup'
abort("restore_item and tar path needs to be passed as arguments to the script") unless ARGV.length == 2
bucket_name = ENV["#{ARGV[0].upcase}_BUCKET_NAME"] || "gitlab-#{ARGV[0]}"
tmp_bucket = ENV['TMP_BUCKET_NAME'] || 'tmp'
backend_type = ENV['BACKUP_BACKEND'] || 's3'
ObjectStorageBackup.new(ARGV[0], ARGV[1], bucket_name, tmp_bucket, backend_type).restore
require 'open3'
require 'fileutils'
class String
def red; "\e[31m#{self}\e[0m" end
def green; "\e[32m#{self}\e[0m" end
def blue; "\e[34m#{self}\e[0m" end
end
class ObjectStorageBackup
attr_accessor :name, :local_tar_path, :remote_bucket_name, :tmp_bucket_name, :backend
def initialize(name, local_tar_path, remote_bucket_name, tmp_bucket_name = 'tmp', backend = 's3')
@name = name
@local_tar_path = local_tar_path
@remote_bucket_name = remote_bucket_name
@tmp_bucket_name = tmp_bucket_name
@backend = backend
end
def backup
if @backend == "s3"
# Check bucket existence by listing, limit 1 to optimize
check_bucket_cmd = %W(s3cmd --limit=1 ls s3://#{@remote_bucket_name})
cmd = %W(s3cmd --stop-on-error --delete-removed sync s3://#{@remote_bucket_name}/ /srv/gitlab/tmp/#{@name}/)
elsif @backend == "gcs"
check_bucket_cmd = %W(gsutil ls gs://#{@remote_bucket_name})
cmd = %W(gsutil -m rsync -r gs://#{@remote_bucket_name} /srv/gitlab/tmp/#{@name})
end
# Check if the bucket exists
output, status = run_cmd(check_bucket_cmd)
unless status.zero?
puts "Bucket not found: #{@remote_bucket_name}. Skipping backup of #{@name} ...".blue
return
end
puts "Dumping #{@name} ...".blue
# create the destination: gsutils requires it to exist, s3cmd does not
FileUtils.mkdir_p("/srv/gitlab/tmp/#{@name}", mode: 0700)
output, status = run_cmd(cmd)
failure_abort('creation of working directory', output) unless status.zero?
# check the destiation for contents. Bucket may have been empty.
if Dir.empty? "/srv/gitlab/tmp/#{@name}"
puts "empty".green
return
end
# build gzip command used for tar compression
gzip_cmd = 'gzip' + (ENV['GZIP_RSYNCABLE'] == 'yes' ? ' --rsyncable' : '')
cmd = %W(tar -cf #{@local_tar_path} -I #{gzip_cmd} -C /srv/gitlab/tmp/#{@name} . )
output, status = run_cmd(cmd)
failure_abort('archive', output) unless status.zero?
puts "done".green
end
def restore
puts "Restoring #{@name} ...".blue
backup_existing
cleanup
restore_from_backup
puts "done".green
end
def failure_abort(action, error_message)
puts "[Error] #{error_message}".red
abort "#{action} of #{@name} failed"
end
def upload_to_object_storage(source_path)
dir_name = File.basename(source_path)
if @backend == "s3"
cmd = %W(s3cmd --stop-on-error sync #{source_path}/ s3://#{@remote_bucket_name}/#{dir_name}/)
elsif @backend == "gcs"
cmd = %W(gsutil -m rsync -r #{source_path}/ gs://#{@remote_bucket_name}/#{dir_name})
end
output, status = run_cmd(cmd)
failure_abort('upload', output) unless status.zero?
end
def backup_existing
backup_file_name = "#{@name}.#{Time.now.to_i}"
if @backend == "s3"
cmd = %W(s3cmd sync s3://#{@remote_bucket_name} s3://#{@tmp_bucket_name}/#{backup_file_name}/)
elsif @backend == "gcs"
cmd = %W(gsutil -m rsync -r gs://#{@remote_bucket_name} gs://#{@tmp_bucket_name}/#{backup_file_name}/)
end
output, status = run_cmd(cmd)
failure_abort('sync existing', output) unless status.zero?
end
def cleanup
if @backend == "s3"
cmd = %W(s3cmd --stop-on-error del --force --recursive s3://#{@remote_bucket_name})
elsif @backend == "gcs"
# Check if the bucket has any objects
list_objects_cmd = %W(gsutil ls gs://#{@remote_bucket_name}/)
output, status = run_cmd(list_objects_cmd)
failure_abort('GCS ls', output) unless status.zero?
# There are no objects in the bucket so skip the cleanup
if output.length == 0
return
end
cmd = %W(gsutil rm -f -r gs://#{@remote_bucket_name}/*)
end
output, status = run_cmd(cmd)
failure_abort('bucket cleanup', output) unless status.zero?
end
def restore_from_backup
extracted_tar_path = File.join(File.dirname(@local_tar_path), "/srv/gitlab/tmp/#{@name}")
FileUtils.mkdir_p(extracted_tar_path, mode: 0700)
failure_abort('restore', "#{@local_tar_path} not found") unless File.exist?(@local_tar_path)
untar_cmd = %W(tar -xf #{@local_tar_path} -C #{extracted_tar_path})
output, status = run_cmd(untar_cmd)
failure_abort('un-archive', output) unless status.zero?
Dir.glob("#{extracted_tar_path}/*").each do |file|
upload_to_object_storage(file)
end
end
def run_cmd(cmd)
_, stdout, wait_thr = Open3.popen2e(*cmd)
return stdout.read, wait_thr.value.exitstatus
end
end