Commit 55b45efb authored by GitLab Bot's avatar GitLab Bot
Browse files

Add latest changes from gitlab-org/gitlab@master

parent 11aec7a48455
......@@ -245,7 +245,7 @@
.select2-highlighted {
.group-result {
.group-path {
color: $white-light;
color: $gray-800;
}
}
}
......
......@@ -118,8 +118,6 @@ def persisted_environment
scope :eager_load_job_artifacts, -> { includes(:job_artifacts) }
scope :with_artifacts_stored_locally, -> { with_existing_job_artifacts(Ci::JobArtifact.archive.with_files_stored_locally) }
scope :with_archived_trace_stored_locally, -> { with_existing_job_artifacts(Ci::JobArtifact.trace.with_files_stored_locally) }
scope :with_artifacts_not_expired, ->() { with_artifacts_archive.where('artifacts_expire_at IS NULL OR artifacts_expire_at > ?', Time.now) }
scope :with_expired_artifacts, ->() { with_artifacts_archive.where('artifacts_expire_at < ?', Time.now) }
scope :last_month, ->() { where('created_at > ?', Date.today - 1.month) }
......
......@@ -65,6 +65,7 @@ class JobArtifact < ApplicationRecord
after_save :update_file_store, if: :saved_change_to_file?
scope :with_files_stored_locally, -> { where(file_store: [nil, ::JobArtifactUploader::Store::LOCAL]) }
scope :with_files_stored_remotely, -> { where(file_store: ::JobArtifactUploader::Store::REMOTE) }
scope :with_file_types, -> (file_types) do
types = self.file_types.select { |file_type| file_types.include?(file_type) }.values
......
......@@ -39,7 +39,7 @@ class Cluster < ApplicationRecord
def self.has_one_cluster_application(name) # rubocop:disable Naming/PredicateName
application = APPLICATIONS[name.to_s]
has_one application.association_name, class_name: application.to_s # rubocop:disable Rails/ReflectionClassName
has_one application.association_name, class_name: application.to_s, inverse_of: :cluster # rubocop:disable Rails/ReflectionClassName
end
has_one_cluster_application :helm
......
......@@ -9,6 +9,7 @@ class LfsObject < ApplicationRecord
has_many :projects, -> { distinct }, through: :lfs_objects_projects
scope :with_files_stored_locally, -> { where(file_store: LfsObjectUploader::Store::LOCAL) }
scope :with_files_stored_remotely, -> { where(file_store: LfsObjectUploader::Store::REMOTE) }
validates :oid, presence: true, uniqueness: true
......
---
title: Increase color contrast of select option path
merge_request:
author:
type: other
---
title: "[ObjectStorage] Allow migrating back to local storage"
merge_request: 16868
author:
type: added
---
title: Set name and updated_at properly in GitHub ReleaseImporter
merge_request: 17020
author:
type: fixed
......@@ -92,7 +92,7 @@ Use an object storage option like AWS S3 to store job artifacts.
DANGER: **Danger:**
If you're enabling S3 in [GitLab HA](high_availability/README.md), you will need to have an [NFS mount set up for CI traces and artifacts](high_availability/nfs.md#a-single-nfs-mount) or enable [live tracing](job_traces.md#new-live-trace-architecture). If these settings are not set, you will risk job traces disappearing or not being saved.
### Object Storage Settings
#### Object Storage Settings
For source installations the following settings are nested under `artifacts:` and then `object_store:`. On Omnibus GitLab installs they are prefixed by `artifacts_object_store_`.
......@@ -105,7 +105,7 @@ For source installations the following settings are nested under `artifacts:` an
| `proxy_download` | Set to true to enable proxying all files served. Option allows to reduce egress traffic as this allows clients to download directly from remote storage instead of proxying all data | `false` |
| `connection` | Various connection options described below | |
#### S3 compatible connection settings
##### S3 compatible connection settings
The connection settings match those provided by [Fog](https://github.com/fog), and are as follows:
......@@ -188,6 +188,14 @@ _The artifacts are stored by default in
sudo -u git -H bundle exec rake gitlab:artifacts:migrate RAILS_ENV=production
```
### Migrating from object storage to local storage
In order to migrate back to local storage:
1. Set both `direct_upload` and `background_upload` to false under the artifacts object storage settings. Don't forget to restart GitLab.
1. Run `rake gitlab:artifacts:migrate_to_local` on your console.
1. Disable `object_storage` for artifacts in `gitlab.rb`. Remember to restart GitLab afterwards.
## Expiring artifacts
If an expiry date is used for the artifacts, they are marked for deletion
......
......@@ -59,46 +59,6 @@ job traces are automatically migrated to it along with the other job artifacts.
See "Phase 4: uploading" in [Data flow](#data-flow) to learn about the process.
## How to archive legacy job trace files
Legacy job traces, which were created before GitLab 10.5, were not archived regularly.
It's the same state with the "2: overwriting" in the above [Data flow](#data-flow).
To archive those legacy job traces, please follow the instruction below.
1. Execute the following command
```bash
gitlab-rake gitlab:traces:archive
```
After you executed this task, GitLab instance queues up Sidekiq jobs (asynchronous processes)
for migrating job trace files from local storage to object storage.
It could take time to complete the all migration jobs. You can check the progress by the following command
```bash
sudo gitlab-rails console
```
```bash
[1] pry(main)> Sidekiq::Stats.new.queues['pipeline_background:archive_trace']
=> 100
```
If the count becomes zero, the archiving processes are done
## How to migrate archived job traces to object storage
> [Introduced][ce-21193] in GitLab 11.3.
If job traces have already been archived into local storage, and you want to migrate those traces to object storage, please follow the instruction below.
1. Ensure [Object storage integration for Job Artifacts](job_artifacts.md#object-storage-settings) is enabled
1. Execute the following command
```bash
gitlab-rake gitlab:traces:migrate
```
## How to remove job traces
There isn't a way to automatically expire old job logs, but it's safe to remove
......
......@@ -61,6 +61,9 @@ To enable external storage of merge request diffs, follow the instructions below
## Using object storage
CAUTION: **WARNING:**
Currently migrating to object storage is **non-reversible**
Instead of storing the external diffs on disk, we recommended the use of an object
store like AWS S3 instead. This configuration relies on valid AWS credentials to
be configured already.
......
......@@ -113,3 +113,39 @@ To migrate all uploads created by legacy uploaders, run:
```shell
bundle exec rake gitlab:uploads:legacy:migrate
```
## Migrate from object storage to local storage
If you need to disable Object Storage for any reason, first you need to migrate
your data out of Object Storage and back into your local storage.
**Before proceeding, it is important to disable both `direct_upload` and `background_upload` under `uploads` settings in `gitlab.rb`**
CAUTION: **Warning:**
**Extended downtime is required** so no new files are created in object storage during
the migration. A configuration setting will be added soon to allow migrating
from object storage to local files with only a brief moment of downtime for configuration changes.
See issue [gitlab-org/gitlab-ce#66144](https://gitlab.com/gitlab-org/gitlab-ce/issues/66144)
### All-in-one rake task
GitLab provides a wrapper rake task that migrates all uploaded files - avatars,
logos, attachments, favicon, etc. - to local storage in one go. Under the hood,
it invokes individual rake tasks to migrate files falling under each of this
category one by one. For details on these rake tasks please [refer to the section above](#individual-rake-tasks),
keeping in mind the task name in this case is `gitlab:uploads:migrate_to_local`.
**Omnibus Installation**
```bash
gitlab-rake "gitlab:uploads:migrate_to_local:all"
```
**Source Installation**
```bash
sudo RAILS_ENV=production -u git -H bundle exec rake gitlab:uploads:migrate_to_local:all
```
After this is done, you may disable Object Storage by undoing the changes described
in the instructions to [configure object storage](../../uploads.md#using-object-storage-core-only)
......@@ -218,6 +218,14 @@ For source installations the settings are nested under `lfs:` and then
will be forwarded to object storage unless `background_upload` is set to
false.
### Migrating back to local storage
In order to migrate back to local storage:
1. Set both `direct_upload` and `background_upload` to false under the LFS object storage settings. Don't forget to restart GitLab.
1. Run `rake gitlab:lfs:migrate_to_local` on your console.
1. Disable `object_storage` for LFS objects in `gitlab.rb`. Remember to restart GitLab afterwards.
## Storage statistics
You can see the total storage used for LFS objects on groups and projects
......
# frozen_string_literal: true
module Gitlab
module Artifacts
class MigrationHelper
def migrate_to_remote_storage(&block)
artifacts = ::Ci::JobArtifact.with_files_stored_locally
migrate(artifacts, ObjectStorage::Store::REMOTE, &block)
end
def migrate_to_local_storage(&block)
artifacts = ::Ci::JobArtifact.with_files_stored_remotely
migrate(artifacts, ObjectStorage::Store::LOCAL, &block)
end
private
def batch_size
ENV.fetch('MIGRATION_BATCH_SIZE', 10).to_i
end
def migrate(artifacts, store, &block)
artifacts.find_each(batch_size: batch_size) do |artifact| # rubocop:disable CodeReuse/ActiveRecord
artifact.file.migrate!(store)
yield artifact if block
rescue => e
raise StandardError.new("Failed to transfer artifact of type #{artifact.file_type} and ID #{artifact.id} with error: #{e.message}")
end
end
end
end
end
......@@ -32,10 +32,11 @@ def already_imported?(release)
def build(release)
{
name: release.name,
tag: release.tag_name,
description: description_for(release),
created_at: release.created_at,
updated_at: release.updated_at,
updated_at: release.created_at,
released_at: release.published_at,
project_id: project.id
}
......
# frozen_string_literal: true
module Gitlab
module Uploads
class MigrationHelper
attr_reader :logger
CATEGORIES = [%w(AvatarUploader Project :avatar),
%w(AvatarUploader Group :avatar),
%w(AvatarUploader User :avatar),
%w(AttachmentUploader Note :attachment),
%w(AttachmentUploader Appearance :logo),
%w(AttachmentUploader Appearance :header_logo),
%w(FaviconUploader Appearance :favicon),
%w(FileUploader Project),
%w(PersonalFileUploader Snippet),
%w(NamespaceFileUploader Snippet),
%w(FileUploader MergeRequest)].freeze
def initialize(args, logger)
prepare_variables(args, logger)
end
def migrate_to_remote_storage
@to_store = ObjectStorage::Store::REMOTE
uploads.each_batch(of: batch_size, &method(:enqueue_batch))
end
def migrate_to_local_storage
@to_store = ObjectStorage::Store::LOCAL
uploads(ObjectStorage::Store::REMOTE).each_batch(of: batch_size, &method(:enqueue_batch))
end
private
def batch_size
ENV.fetch('MIGRATION_BATCH_SIZE', 200).to_i
end
def prepare_variables(args, logger)
@mounted_as = args.mounted_as&.gsub(':', '')&.to_sym
@uploader_class = args.uploader_class.constantize
@model_class = args.model_class.constantize
@logger = logger
end
def enqueue_batch(batch, index)
job = ObjectStorage::MigrateUploadsWorker.enqueue!(batch,
@model_class,
@mounted_as,
@to_store)
logger.info(message: "[Uploads migration] Enqueued upload migration job", index: index, job_id: job)
rescue ObjectStorage::MigrateUploadsWorker::SanityCheckError => e
# continue for the next batch
logger.warn(message: "[Uploads migration] Could not enqueue batch", ids: batch.ids, reason: e.message) # rubocop:disable CodeReuse/ActiveRecord
end
# rubocop:disable CodeReuse/ActiveRecord
def uploads(store_type = [nil, ObjectStorage::Store::LOCAL])
Upload.class_eval { include EachBatch } unless Upload < EachBatch
Upload
.where(store: store_type,
uploader: @uploader_class.to_s,
model_type: @model_class.base_class.sti_name)
end
# rubocop:enable CodeReuse/ActiveRecord
end
end
end
......@@ -6,18 +6,31 @@ namespace :gitlab do
namespace :artifacts do
task migrate: :environment do
logger = Logger.new(STDOUT)
logger.info('Starting transfer of artifacts')
logger.info('Starting transfer of artifacts to remote storage')
Ci::Build.joins(:project)
.with_artifacts_stored_locally
.find_each(batch_size: 10) do |build|
helper = Gitlab::Artifacts::MigrationHelper.new
build.artifacts_file.migrate!(ObjectStorage::Store::REMOTE)
build.artifacts_metadata.migrate!(ObjectStorage::Store::REMOTE)
begin
helper.migrate_to_remote_storage do |artifact|
logger.info("Transferred artifact ID #{artifact.id} of type #{artifact.file_type} with size #{artifact.size} to object storage")
end
rescue => e
logger.error(e.message)
end
end
task migrate_to_local: :environment do
logger = Logger.new(STDOUT)
logger.info('Starting transfer of artifacts to local storage')
helper = Gitlab::Artifacts::MigrationHelper.new
logger.info("Transferred artifact ID #{build.id} with size #{build.artifacts_size} to object storage")
begin
helper.migrate_to_local_storage do |artifact|
logger.info("Transferred artifact ID #{artifact.id} of type #{artifact.file_type} with size #{artifact.size} to local storage")
end
rescue => e
logger.error("Failed to transfer artifacts of #{build.id} with error: #{e.message}")
logger.error(e.message)
end
end
end
......
......@@ -17,5 +17,20 @@ namespace :gitlab do
logger.error("Failed to transfer LFS object #{lfs_object.oid} with error: #{e.message}")
end
end
task migrate_to_local: :environment do
logger = Logger.new(STDOUT)
logger.info('Starting transfer of LFS files to local storage')
LfsObject.with_files_stored_remotely
.find_each(batch_size: 10) do |lfs_object|
lfs_object.file.migrate!(LfsObjectUploader::Store::LOCAL)
logger.info("Transferred LFS object #{lfs_object.oid} of size #{lfs_object.size.to_i.bytes} to local storage")
rescue => e
logger.error("Failed to transfer LFS object #{lfs_object.oid} with error: #{e.message}")
end
end
end
end
require 'logger'
require 'resolv-replace'
desc "GitLab | Archive legacy traces to trace artifacts"
namespace :gitlab do
namespace :traces do
task archive: :environment do
logger = Logger.new(STDOUT)
logger.info('Archiving legacy traces')
Ci::Build.finished.without_archived_trace
.order(id: :asc)
.find_in_batches(batch_size: 1000) do |jobs|
job_ids = jobs.map { |job| [job.id] }
ArchiveTraceWorker.bulk_perform_async(job_ids)
logger.info("Scheduled #{job_ids.count} jobs. From #{job_ids.min} to #{job_ids.max}")
end
end
task migrate: :environment do
logger = Logger.new(STDOUT)
logger.info('Starting transfer of job traces')
Ci::Build.joins(:project)
.with_archived_trace_stored_locally
.find_each(batch_size: 10) do |build|
build.job_artifacts_trace.file.migrate!(ObjectStorage::Store::REMOTE)
logger.info("Transferred job trace of #{build.id} to object storage")
rescue => e
logger.error("Failed to transfer artifacts of #{build.id} with error: #{e.message}")
end
end
end
end
......@@ -3,19 +3,7 @@ namespace :gitlab do
namespace :migrate do
desc "GitLab | Uploads | Migrate all uploaded files to object storage"
task all: :environment do
categories = [%w(AvatarUploader Project :avatar),
%w(AvatarUploader Group :avatar),
%w(AvatarUploader User :avatar),
%w(AttachmentUploader Note :attachment),
%w(AttachmentUploader Appearance :logo),
%w(AttachmentUploader Appearance :header_logo),
%w(FaviconUploader Appearance :favicon),
%w(FileUploader Project),
%w(PersonalFileUploader Snippet),
%w(NamespaceFileUploader Snippet),
%w(FileUploader MergeRequest)]
categories.each do |args|
Gitlab::Uploads::MigrationHelper::CATEGORIES.each do |args|
Rake::Task["gitlab:uploads:migrate"].invoke(*args)
Rake::Task["gitlab:uploads:migrate"].reenable
end
......@@ -25,34 +13,23 @@ namespace :gitlab do
# The following is the actual rake task that migrates uploads of specified
# category to object storage
desc 'GitLab | Uploads | Migrate the uploaded files of specified type to object storage'
task :migrate, [:uploader_class, :model_class, :mounted_as] => :environment do |task, args|
batch_size = ENV.fetch('BATCH', 200).to_i
@to_store = ObjectStorage::Store::REMOTE
@mounted_as = args.mounted_as&.gsub(':', '')&.to_sym
@uploader_class = args.uploader_class.constantize
@model_class = args.model_class.constantize
uploads.each_batch(of: batch_size, &method(:enqueue_batch))
task :migrate, [:uploader_class, :model_class, :mounted_as] => :environment do |_t, args|
Gitlab::Uploads::MigrationHelper.new(args, Logger.new(STDOUT)).migrate_to_remote_storage
end
def enqueue_batch(batch, index)
job = ObjectStorage::MigrateUploadsWorker.enqueue!(batch,
@model_class,
@mounted_as,
@to_store)
puts "Enqueued job ##{index}: #{job}"
rescue ObjectStorage::MigrateUploadsWorker::SanityCheckError => e
# continue for the next batch
puts "Could not enqueue batch (#{batch.ids}) #{e.message}".color(:red)
namespace :migrate_to_local do
desc "GitLab | Uploads | Migrate all uploaded files to local storage"
task all: :environment do
Gitlab::Uploads::MigrationHelper::CATEGORIES.each do |args|
Rake::Task["gitlab:uploads:migrate_to_local"].invoke(*args)
Rake::Task["gitlab:uploads:migrate_to_local"].reenable
end
end
end
def uploads
Upload.class_eval { include EachBatch } unless Upload < EachBatch
Upload
.where(store: [nil, ObjectStorage::Store::LOCAL],
uploader: @uploader_class.to_s,
model_type: @model_class.base_class.sti_name)
desc 'GitLab | Uploads | Migrate the uploaded files of specified type to local storage'
task :migrate_to_local, [:uploader_class, :model_class, :mounted_as] => :environment do |_t, args|
Gitlab::Uploads::MigrationHelper.new(args, Logger.new(STDOUT)).migrate_to_local_storage
end
end
end
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment