doc/administration/lfs/_index.md
{{< details >}}
{{< /details >}}
Use Git Large File Storage (LFS) to store large files in a Git repository without increasing its size or affecting performance. You can enable or disable LFS, configure local or remote storage for LFS objects, and migrate objects between storage types.
For user documentation, see Git Large File Storage (LFS).
Prerequisites:
LFS is enabled by default. To disable it:
{{< tabs >}}
{{< tab title="Linux package (Omnibus)" >}}
Edit /etc/gitlab/gitlab.rb:
# Change to true to enable lfs - enabled by default if not defined
gitlab_rails['lfs_enabled'] = false
Save the file and reconfigure GitLab:
sudo gitlab-ctl reconfigure
{{< /tab >}}
{{< tab title="Helm chart (Kubernetes)" >}}
Export the Helm values:
helm get values gitlab > gitlab_values.yaml
Edit gitlab_values.yaml:
global:
appConfig:
lfs:
enabled: false
Save the file and apply the new values:
helm upgrade -f gitlab_values.yaml gitlab gitlab/gitlab
{{< /tab >}}
{{< tab title="Docker" >}}
Edit docker-compose.yml:
version: "3.6"
services:
gitlab:
environment:
GITLAB_OMNIBUS_CONFIG: |
gitlab_rails['lfs_enabled'] = false
Save the file and restart GitLab:
docker compose up -d
{{< /tab >}}
{{< tab title="Self-compiled (source)" >}}
Edit /home/git/gitlab/config/gitlab.yml:
production: &base
lfs:
enabled: false
Save the file and restart GitLab:
# For systems running systemd
sudo systemctl restart gitlab.target
# For systems running SysV init
sudo service gitlab restart
{{< /tab >}}
{{< /tabs >}}
Git LFS objects can be large in size. By default, they are stored on the server GitLab is installed on.
[!note] For Docker installations, you can change the path where your data is mounted. For the Helm chart, use object storage.
To change the default local storage path location:
{{< tabs >}}
{{< tab title="Linux package (Omnibus)" >}}
Edit /etc/gitlab/gitlab.rb:
# /var/opt/gitlab/gitlab-rails/shared/lfs-objects by default.
gitlab_rails['lfs_storage_path'] = "/mnt/storage/lfs-objects"
Save the file and reconfigure GitLab:
sudo gitlab-ctl reconfigure
{{< /tab >}}
{{< tab title="Self-compiled (source)" >}}
Edit /home/git/gitlab/config/gitlab.yml:
# /home/git/gitlab/shared/lfs-objects by default.
production: &base
lfs:
storage_path: /mnt/storage/lfs-objects
Save the file and restart GitLab:
# For systems running systemd
sudo systemctl restart gitlab.target
# For systems running SysV init
sudo service gitlab restart
{{< /tab >}}
{{< /tabs >}}
You can store LFS objects in remote object storage. This allows you to reduce reads and writes to the local disk, and free up disk space significantly.
You should use the consolidated object storage settings.
You can migrate the LFS objects from local storage to object storage. The processing is done in the background and requires no downtime.
Migrate the LFS objects:
{{< tabs >}}
{{< tab title="Linux package (Omnibus)" >}}
sudo gitlab-rake gitlab:lfs:migrate
{{< /tab >}}
{{< tab title="Docker" >}}
sudo docker exec -t <container name> gitlab-rake gitlab:lfs:migrate
{{< /tab >}}
{{< tab title="Self-compiled (source)" >}}
sudo -u git -H bundle exec rake gitlab:lfs:migrate RAILS_ENV=production
{{< /tab >}}
{{< /tabs >}}
Optional. Track the progress and verify that all job LFS objects migrated successfully using the PostgreSQL console.
Open a PostgreSQL console:
{{< tabs >}}
{{< tab title="Linux package (Omnibus)" >}}
sudo gitlab-psql
{{< /tab >}}
{{< tab title="Docker" >}}
sudo docker exec -it <container_name> /bin/bash
gitlab-psql
{{< /tab >}}
{{< tab title="Self-compiled (source)" >}}
sudo -u git -H psql -d gitlabhq_production
{{< /tab >}}
{{< /tabs >}}
Verify that all LFS files migrated to object storage with the following
SQL query. The number of objectstg should be the same as total:
gitlabhq_production=# SELECT count(*) AS total, sum(case when file_store = '1' then 1 else 0 end) AS filesystem, sum(case when file_store = '2' then 1 else 0 end) AS objectstg FROM lfs_objects;
total | filesystem | objectstg
------+------------+-----------
2409 | 0 | 2409
Verify that there are no files on disk in the lfs-objects directory:
{{< tabs >}}
{{< tab title="Linux package (Omnibus)" >}}
sudo find /var/opt/gitlab/gitlab-rails/shared/lfs-objects -type f | grep -v tmp | wc -l
{{< /tab >}}
{{< tab title="Docker" >}}
Assuming you mounted /var/opt/gitlab to /srv/gitlab:
sudo find /srv/gitlab/gitlab-rails/shared/lfs-objects -type f | grep -v tmp | wc -l
{{< /tab >}}
{{< tab title="Self-compiled (source)" >}}
sudo find /home/git/gitlab/shared/lfs-objects -type f | grep -v tmp | wc -l
{{< /tab >}}
{{< /tabs >}}
[!note] For the Helm chart, you should use object storage.
To migrate back to local storage:
{{< tabs >}}
{{< tab title="Linux package (Omnibus)" >}}
Migrate the LFS objects:
sudo gitlab-rake gitlab:lfs:migrate_to_local
Edit /etc/gitlab/gitlab.rb and
disable object storage
for LFS objects:
gitlab_rails['object_store']['objects']['lfs']['enabled'] = false
Save the file and reconfigure GitLab:
sudo gitlab-ctl reconfigure
{{< /tab >}}
{{< tab title="Docker" >}}
Migrate the LFS objects:
sudo docker exec -t <container name> gitlab-rake gitlab:lfs:migrate_to_local
Edit docker-compose.yml and disable object storage for LFS objects:
version: "3.6"
services:
gitlab:
environment:
GITLAB_OMNIBUS_CONFIG: |
gitlab_rails['object_store']['objects']['lfs']['enabled'] = false
Save the file and restart GitLab:
docker compose up -d
{{< /tab >}}
{{< tab title="Self-compiled (source)" >}}
Migrate the LFS objects:
sudo -u git -H bundle exec rake gitlab:lfs:migrate_to_local RAILS_ENV=production
Edit /home/git/gitlab/config/gitlab.yml and disable object storage for LFS objects:
production: &base
object_store:
objects:
lfs:
enabled: false
Save the file and restart GitLab:
# For systems running systemd
sudo systemctl restart gitlab.target
# For systems running SysV init
sudo service gitlab restart
{{< /tab >}}
{{< /tabs >}}
{{< history >}}
{{< /history >}}
[!warning] This feature is affected by a known issue (resolved in Git LFS 3.6.0). If you clone a repository with multiple Git LFS objects using the pure SSH protocol, the client might crash due to a
nilpointer reference.
git-lfs 3.0.0
released support for using SSH as the transfer protocol instead of HTTP.
SSH is handled transparently by the git-lfs command line tool.
When pure SSH protocol support is enabled and git is configured to use SSH,
all LFS operations happen over SSH. For example, when the Git remote is
[email protected]:gitlab-org/gitlab.git. You can't configure git and git-lfs
to use different protocols. From version 3.0, git-lfs attempts to use the pure
SSH protocol initially and, if support is not enabled or available, it falls back
to using HTTP.
Prerequisites:
git-lfs version must be v3.5.1 or higher.To switch Git LFS to use pure SSH protocol:
{{< tabs >}}
{{< tab title="Linux package (Omnibus)" >}}
Edit /etc/gitlab/gitlab.rb:
gitlab_shell['lfs_pure_ssh_protocol'] = true
Save the file and reconfigure GitLab:
sudo gitlab-ctl reconfigure
{{< /tab >}}
{{< tab title="Helm chart (Kubernetes)" >}}
Export the Helm values:
helm get values gitlab > gitlab_values.yaml
Edit gitlab_values.yaml:
gitlab:
gitlab-shell:
config:
lfs:
pureSSHProtocol: true
Save the file and apply the new values:
helm upgrade -f gitlab_values.yaml gitlab gitlab/gitlab
{{< /tab >}}
{{< tab title="Docker" >}}
Edit docker-compose.yml:
services:
gitlab:
environment:
GITLAB_OMNIBUS_CONFIG: |
gitlab_shell['lfs_pure_ssh_protocol'] = true
Save the file and restart GitLab and its services:
docker compose up -d
{{< /tab >}}
{{< tab title="Self-compiled (source)" >}}
Edit /home/git/gitlab-shell/config.yml:
lfs:
pure_ssh_protocol: true
Save the file and restart GitLab Shell:
# For systems running systemd
sudo systemctl restart gitlab-shell.target
# For systems running SysV init
sudo service gitlab-shell restart
{{< /tab >}}
{{< /tabs >}}
You can see the total storage used for LFS objects for groups and projects in:
[!note] The storage statistics count each LFS object for every project linking to it.
An error about a missing LFS object may occur in either of these situations:
When migrating LFS objects from disk to object storage, with error messages like:
ERROR -- : Failed to transfer LFS object
006622269c61b41bf14a22bbe0e43be3acf86a4a446afb4250c3794ea47541a7
with error: No such file or directory @ rb_sysopen -
/var/opt/gitlab/gitlab-rails/shared/lfs-objects/00/66/22269c61b41bf14a22bbe0e43be3acf86a4a446afb4250c3794ea47541a7
(Line breaks have been added for legibility.)
When running the
integrity check for LFS objects
with the VERBOSE=1 parameter.
The database can have records for LFS objects which are not on disk. The database entry may prevent a new copy of the object from being pushed. To delete these references:
Query the object that's reported as missing in the rails console, to return a file path:
lfs_object = LfsObject.find_by(oid: '006622269c61b41bf14a22bbe0e43be3acf86a4a446afb4250c3794ea47541a7')
lfs_object.file.path
Check on disk or object storage if it exists:
ls -al /var/opt/gitlab/gitlab-rails/shared/lfs-objects/00/66/22269c61b41bf14a22bbe0e43be3acf86a4a446afb4250c3794ea47541a7
If the file is not present, remove the database records with the Rails console:
# First delete the parent records and then destroy the record itself
lfs_object.lfs_objects_projects.destroy_all
lfs_object.destroy
To remove references to multiple missing LFS objects at once:
Open the GitLab Rails console.
Run the following script:
lfs_files_deleted = 0
LfsObject.find_each do |lfs_file|
next if lfs_file.file.file.exists?
lfs_files_deleted += 1
p "LFS file with ID #{lfs_file.id} and path #{lfs_file.file.path} is missing."
# lfs_file.lfs_objects_projects.destroy_all # Uncomment to delete parent records
# lfs_file.destroy # Uncomment to destroy the LFS object reference
end
p "Count of identified/destroyed invalid references: #{lfs_files_deleted}"
This script identifies all missing LFS objects in the database. Before deleting any records:
If you configure GitLab to disable TLS v1.2 and only enable TLS v1.3 connections, LFS operations require a Git LFS client version 2.11.0 or later. If you use a Git LFS client earlier than version 2.11.0, GitLab displays an error:
batch response: Post https://username:***@gitlab.example.com/tool/releases.git/info/lfs/objects/batch: remote error: tls: protocol version not supported
error: failed to fetch some objects from 'https://username:[MASKED]@gitlab.example.com/tool/releases.git/info/lfs'
When using GitLab CI over a TLS v1.3 configured GitLab server, you must upgrade to GitLab Runner 13.2.0 or later to receive an updated Git LFS client version with the included GitLab Runner Helper image.
To check an installed Git LFS client's version, run this command:
git lfs version
Connection refused errorsIf you push or mirror LFS objects and receive errors like the following:
dial tcp <IP>:443: connect: connection refusedConnection refused - connect(2) for \"<target-or-proxy-IP>\" port 443a firewall or proxy rule may be terminating the connection.
If connection checks with standard Unix tools or manual Git pushes are successful, the rule may be related to the size of the request.
When LFS has been configured with object storage and proxy_download set to
false, you might see an error when previewing a PDF file from the Web browser:
An error occurred while loading the file. Please try again later.
This occurs due to Cross-Origin Resource Sharing (CORS) restrictions: the browser attempts to load the PDF from object storage, but the object storage provider rejects the request because the GitLab domain differs from the object storage domain.
To fix this issue, configure your object storage provider's CORS settings to allow the GitLab domain. See the following documentation for more details:
Forking in progress messageIf you are forking a project with multiple LFS files, the operation might get stuck with a Forking in progress message.
If you encounter this, follow these steps to diagnose and resolve the issue:
Check your exceptions_json.log file for the following error message:
"error_message": "Unable to fork project 12345 for repository
@hashed/11/22/encoded-path -> @hashed/33/44/encoded-new-path:
Source project has too many LFS objects"
This error indicates that you've reached the default limit of 100,000 LFS files, as described in issue 476693.
Increase the value of the GITLAB_LFS_MAX_OID_TO_FETCH variable:
Open the configuration file /etc/gitlab/gitlab.rb.
Add or update the variable:
gitlab_rails['env'] = {
"GITLAB_LFS_MAX_OID_TO_FETCH" => "NEW_VALUE"
}
Replace NEW_VALUE with a number based on your requirements.
Apply the changes. Run:
sudo gitlab-ctl reconfigure
For more information, see reconfigure a Linux package installation.
Repeat the fork operation.
[!note] For the GitLab Helm chart, use
extraEnvto configure the environment variableGITLAB_LFS_MAX_OID_TO_FETCH.