GitLab 17 changes

DETAILS: Tier: Free, Premium, Ultimate Offering: Self-managed

This page contains upgrade information for minor and patch versions of GitLab 17. Ensure you review these instructions for:

Your installation type.
All versions between your current version and your target version.

For more information about upgrading GitLab Helm Chart, see the release notes for 8.0.

Issues to be aware of when upgrading from 16.11

You should migrate to the new runner registration workflow before upgrading to GitLab 17.0.

In GitLab 16.0, we introduced a new runner creation workflow that uses runner authentication tokens to register runners. The legacy workflow that uses registration tokens is now disabled by default in GitLab 17.0 and will be removed in GitLab 18.0. If registration tokens are still being used, upgrading to GitLab 17.0 will cause runner registration to fail.
Gitaly storages can no longer share the same path as in this example:
```
gitaly['configuration'] = {
  storage: [
    {
       name: 'default',
       path: '/var/opt/gitlab/git-data/repositories',
    },
    {
       name: 'duplicate-path',
       path: '/var/opt/gitlab/git-data/repositories',
    },
  ],
}
```
In this example, the duplicate-path storage must be removed or relocated to a new path. If you have more than one Gitaly node, you must ensure only the corresponding storage for that node is listed in that node's gitlab.rb file.

If the storage is removed from a node's gitlab.rb file, then any projects associated with it must have their storage updated in the GitLab database. You can update their storage using the Rails console. For example:
```
$ sudo gitlab-rails console
Project.where(repository_storage: 'duplicate-path').update_all(repository_storage: 'default')
```
Migration failures when upgrading from GitLab 16.x directly to GitLab 17.1.0 or 17.1.1.

Due to a bug in GitLab 17.1.0 and 17.1.1 where a background job completion did not get enforced correctly, there can be failures when upgrading directly to GitLab 17.1.0 and 17.1.1. The error during the migration of the upgrade looks like the following:
```
main: == [advisory_lock_connection] object_id: 55460, pg_backend_pid: 8714
main: == 20240531173207 ValidateNotNullCheckConstraintOnEpicsIssueId: migrating =====
main: -- execute("SET statement_timeout TO 0")
main:    -> 0.0004s
main: -- execute("ALTER TABLE epics VALIDATE CONSTRAINT check_450724d1bb;")
main: -- execute("RESET statement_timeout")
main: == [advisory_lock_connection] object_id: 55460, pg_backend_pid: 8714
STDERR:
```
This issue occurs because the background migration that got introduced in GitLab 17.0 didn't complete. To upgrade, either:
- Upgrade to GitLab 17.0 and wait until all background migrations are completed.
- Upgrade to GitLab 17.1 and then manually execute the background job and the migration by running the following command:
```
sudo gitlab-rake gitlab:background_migrations:finalize[BackfillEpicBasicFieldsToWorkItemRecord,epics,id,'[null]']
```
Now you should be able to complete the migrations in GitLab 17.1 and finish the upgrade. This bug has been fixed with GitLab 17.1.2 and upgrading from GitLab 16.x directly to 17.1.2 will not cause these issues.

Linux package installations

Specific information applies to Linux package installations:

The binaries for PostgreSQL 13 have been removed.

Prior to upgrading, you must ensure your installation is using PostgreSQL 14.

Non-expiring access tokens

Access tokens that have no expiration date are valid indefinitely, which is a security risk if the access token is divulged.

When you upgrade to GitLab 16.0 and later, any personal, project, or group access token that does not have an expiration date automatically has an expiration date set at one year from the date of upgrade.

Before this automatic expiry date is applied, you should do the following to minimize disruption:

For more information, see the:

Issues to be aware of when upgrading from 17.1 and earlier

If the customer is using GitLab Duo and upgrading to GitLab 17.2.3 or earlier, they must do both of the following:
- Resynchronize their license.
- Restart the server after the upgrade.
If the customer is using GitLab Duo and upgrading to GitLab 17.2.4 or later, they must do either of the following:
- Resynchronize their license.
- Wait until the next scheduled license synchronization, which happens every 24 hours.

After the customer has upgraded to GitLab 17.2.4 or later, these steps are not required for future upgrades.

For more information, see issue 480328.

Issues to be aware of when upgrading from 17.3

Migration failures when upgrading from GitLab 17.3.

When upgrading from 17.3 to 17.4, there is a slight chance of encountering an error. During the migration process, you might see an error message like the one below:

main: == [advisory_lock_connection] object_id: 127900, pg_backend_pid: 76263
main: == 20240812040748 AddUniqueConstraintToRemoteDevelopmentAgentConfigs: migrating
main: -- transaction_open?(nil)
main:    -> 0.0000s
main: -- view_exists?(:postgres_partitions)
main:    -> 0.0181s
main: -- index_exists?(:remote_development_agent_configs, :cluster_agent_id, {:name=>"index_remote_development_agent_configs_on_unique_agent_id", :unique=>true, :algorithm=>:concurrently})
main:    -> 0.0026s
main: -- execute("SET statement_timeout TO 0")
main:    -> 0.0004s
main: -- add_index(:remote_development_agent_configs, :cluster_agent_id, {:name=>"index_remote_development_agent_configs_on_unique_agent_id", :unique=>true, :algorithm=>:concurrently})
main: -- execute("RESET statement_timeout")
main:    -> 0.0002s
main: == [advisory_lock_connection] object_id: 127900, pg_backend_pid: 76263
rake aborted!
StandardError: An error has occurred, all later migrations canceled:

PG::UniqueViolation: ERROR:  could not create unique index "index_remote_development_agent_configs_on_unique_agent_id"
DETAIL:  Key (cluster_agent_id)=(1000141) is duplicated.

This error occurs because the migration adds a unique constraint on the cluster_agent_id column in the remote_development_agent_configs table, but there are still duplicate entries. The previous migration is supposed to remove these duplicates, but in rare cases, new duplicates may be inserted between the two migrations.

To safely resolve this issue, follow these steps:

Open the Rails console where the migrations are being run.
Copy and paste the script below into the console and execute it.
Re-run the migrations, and they should complete successfully.

# Get the IDs to keep for each cluster_agent_id; if there are duplicates, only the row with the latest updated_at will be kept.
latest_ids = ::RemoteDevelopment::RemoteDevelopmentAgentConfig.select("DISTINCT ON (cluster_agent_id) id")
  .order("cluster_agent_id, updated_at DESC")
  .map(&:id)

# Get the list of remote_development_agent_configs to be removed.
agent_configs_to_remove = ::RemoteDevelopment::RemoteDevelopmentAgentConfig.where.not(id: latest_ids)

# Delete all duplicated agent_configs.
agent_configs_to_remove.delete_all

17.5.0

With the upgrade to OpenSSL version 3:

GitLab requires TLS 1.2 or higher for all outgoing and incoming TLS connections.
TLS/SSL certificates must have at least 112 bits of security. RSA, DSA, and DH keys shorter than 2048 bits, and ECC keys shorter than 224 bits are prohibited.

Older services, such as LDAP and Webhook servers, may still use TLS 1.1. However, TLS 1.0 and 1.1 have reached end-of-life and are no longer considered secure. GitLab will fail to connect to services using TLS 1.0 or 1.1 with a no protocols available error message.

In addition, OpenSSL 3 increased the default security level from level 1 to 2, raising the number of bits of security from 80 to 112. For example, a certificate signed with an RSA key can use RSA-2048 but not RSA-1024. GitLab will fail to connect to a service that uses a certificate signed with insufficient bits with a certificate key too weak error message.

Check the GitLab documentation on securing your installation for more details.

17.4.0

Starting with GitLab 17.4, new GitLab installations have a different database schema regarding ID columns.
- All previous integer (32 bits) ID columns (for example columns like id, %_id, %_ids) are now created as bigint (64 bits).
- Existing installations will migrate from 32 bit to 64 bit integers in later releases when database migrations ship to perform this change.
- If you are building a new GitLab environment to test upgrades, install GitLab 17.3 or earlier to get the same integer types as your existing environments. You can then upgrade to later releases to run the same database migrations as your existing environments. This isn't necessary if you're restoring from backup into the new environment as the database restore removes the existing database schema definition and uses the definition that's stored as part of the backup.
Git 2.46.0 and later is required by Gitaly. For installations from source, you should use the Git version provided by Gitaly.

17.3.0

Git 2.45.0 and later is required by Gitaly. For installations from source, you should use the Git version provided by Gitaly.

Geo installations

Geo Replication Details pages for a secondary site appear to be empty even if Geo replication is working, see issue 468509. There is no known workaround. The bug is fixed in GitLab 17.4.

Affected releases:

Affected minor releases Affected patch releases Fixed in

17.0 All 17.0.7

17.1 All 17.1.7

17.2 All 17.2.5

17.3 All 17.3.1

Affected minor releases	Affected patch releases	Fixed in
17.0	All	17.0.7
17.1	All	17.1.7
17.2	All	17.2.5
17.3	All	17.3.1

17.2.1

Upgrades to GitLab 17.2.1 can fail because of unknown sequences in the database. This issue has been fixed in GitLab 17.2.2.

Upgrades to GitLab 17.2.1 may fail with the error:

PG::DependentObjectsStillExist: ERROR: cannot drop desired object(s) because other objects depend on them

As described in this issue, this database sequence ownership issue has been fixed in GitLab 17.2.1. However, you might encounter this if the migrations in 17.2.0 did not complete, but the Linux package prevents the upgrade to 17.2.1 or later because of a malformed JSON file. For example, you might see this error:

Malformed configuration JSON file found at /opt/gitlab/embedded/nodes/gitlab.example.com.json.
This usually happens when your last run of `gitlab-ctl reconfigure` didn't complete successfully.
This file is used to check if any of the unsupported configurations are enabled,
and hence require a working reconfigure before upgrading.
Please run `sudo gitlab-ctl reconfigure` to fix it and try again.

The current workaround is to:

Remove the JSON files in /opt/gitlab/embedded/nodes:
```
rm /opt/gitlab/embedded/nodes/*.json
```
Upgrade to GitLab 17.2.1 or higher.

Geo installations

In GitLab 16.11 through GitLab 17.2, a missing PostgreSQL index can cause high CPU usage, slow job artifact verification progress, and slow or timed out Geo metrics status updates. The index was added in GitLab 17.3. To manually add the index, see Geo Troubleshooting - High CPU usage on primary during job artifact verification.

Affected releases:

Affected minor releases Affected patch releases Fixed in

16.11 All None

17.0 All 17.0.7

17.1 All 17.1.7

17.2 All 17.2.5
Geo Replication Details pages for a secondary site appear to be empty even if Geo replication is working, see issue 468509. There is no known workaround. The bug is fixed in GitLab 17.4.

Affected releases:

Affected minor releases Affected patch releases Fixed in

17.0 All 17.0.7

17.1 All 17.1.7

17.2 All 17.2.5

17.3 All 17.3.1

Affected minor releases	Affected patch releases	Fixed in
16.11	All	None
17.0	All	17.0.7
17.1	All	17.1.7
17.2	All	17.2.5

Affected minor releases	Affected patch releases	Fixed in
17.0	All	17.0.7
17.1	All	17.1.7
17.2	All	17.2.5
17.3	All	17.3.1

17.1.0

Bitbucket identities with untrusted extern_uid are deleted. For more information, see issue 452426.
The default changelog template generates links as full URLs instead of GitLab specific references. For more information, see merge request 155806.
Git 2.44.0 and later is required by Gitaly. For self-compiled installations, you should use the Git version provided by Gitaly.
Upgrading to GitLab 17.1.0 or 17.1.1 or having unfinished background migrations from GitLab 17.0 can result in a failure when running the migrations. This is due to a bug. Issue 468875 has been fixed with GitLab 17.1.2.

Geo installations

In GitLab 16.11 through GitLab 17.2, a missing PostgreSQL index can cause high CPU usage, slow job artifact verification progress, and slow or timed out Geo metrics status updates. The index was added in GitLab 17.3. To manually add the index, see Geo Troubleshooting - High CPU usage on primary during job artifact verification.

Affected releases:

Affected minor releases Affected patch releases Fixed in

16.11 All None

17.0 All 17.0.7

17.1 All 17.1.7

17.2 All 17.2.5
Geo Replication Details pages for a secondary site appear to be empty even if Geo replication is working, see issue 468509. There is no known workaround. The bug is fixed in GitLab 17.4.

Affected releases:

Affected minor releases Affected patch releases Fixed in

17.0 All 17.0.7

17.1 All 17.1.7

17.2 All 17.2.5

17.3 All 17.3.1

Affected minor releases	Affected patch releases	Fixed in
16.11	All	None
17.0	All	17.0.7
17.1	All	17.1.7
17.2	All	17.2.5

Affected minor releases	Affected patch releases	Fixed in
17.0	All	17.0.7
17.1	All	17.1.7
17.2	All	17.2.5
17.3	All	17.3.1

17.0.0

Geo installations

In GitLab 16.11 through GitLab 17.2, a missing PostgreSQL index can cause high CPU usage, slow job artifact verification progress, and slow or timed out Geo metrics status updates. The index was added in GitLab 17.3. To manually add the index, see Geo Troubleshooting - High CPU usage on primary during job artifact verification.

Affected releases:

Affected minor releases Affected patch releases Fixed in

16.11 All None

17.0 All 17.0.7

17.1 All 17.1.7

17.2 All 17.2.5
Geo Replication Details pages for a secondary site appear to be empty even if Geo replication is working, see issue 468509. There is no known workaround. The bug is fixed in GitLab 17.4.

Affected releases:

Affected minor releases Affected patch releases Fixed in

17.0 All 17.0.7

17.1 All 17.1.7

17.2 All 17.2.5

17.3 All 17.3.1