Skip to content

Geo sites Admin area

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

You can configure various settings for GitLab Geo sites. For more information, see Geo documentation.

On either the primary or secondary site:

  1. On the left sidebar, at the bottom, select Admin area.
  2. Select Geo > Sites.

Common settings

All Geo sites have the following settings:

Setting Description
Primary This marks a Geo site as primary site. There can be only one primary site.
Name The unique identifier for the Geo site. It's highly recommended to use a physical location as a name. Good examples are "London Office" or "us-east-1". Avoid words like "primary", "secondary", "Geo", or "DR". This makes the failover process easier because the physical location does not change, but the Geo site role can. All nodes in a single Geo site use the same site name. Nodes use the gitlab_rails['geo_node_name'] setting in /etc/gitlab/gitlab.rb to lookup their Geo site record in the PostgreSQL database. If gitlab_rails['geo_node_name'] is not set, the node's external_url with trailing slash is used as fallback. The value of Name is case-sensitive, and most characters are allowed.
URL The instance's user-facing URL.

The site you're currently browsing is indicated with a blue Current label, and the primary node is listed first as Primary site.

Secondary site settings

Secondary sites have a number of additional settings available:

Setting Description
Selective synchronization Enable Geo selective sync for this secondary site.
Repository sync capacity Number of concurrent requests this secondary site makes to the primary site when backfilling repositories.
File sync capacity Number of concurrent requests this secondary site makes to the primary site when backfilling files.

Geo backfill

Secondary sites are notified of changes to repositories and files by the primary site, and always attempt to synchronize those changes as quickly as possible.

Backfill is the act of populating the secondary site with repositories and files that existed before the secondary site was added to the database. Because there may be extremely large numbers of repositories and files, it's not feasible to attempt to download them all at once; so, GitLab places an upper limit on the concurrency of these operations.

How long the backfill takes is dependent on the maximum concurrency, but higher values place more strain on the primary site. The limits are configurable. If your primary site has lots of surplus capacity, you can increase the values to complete backfill in a shorter time. If it's under heavy load and backfill reduces its availability for standard requests, you can decrease them.

Set up the internal URLs

You can set up a different URL for synchronization between the primary and secondary site.

The primary site's Internal URL is used by secondary sites to contact it (to sync repositories, for example). The name Internal URL distinguishes it from External URL, which is used by users. Internal URL does not need to be a private address.

When Geo secondary proxying is enabled, the primary uses the secondary's internal URL to contact it directly.

The internal URL defaults to external URL. To change it:

  1. On the left sidebar, at the bottom, select Admin area.
  2. Select Geo > Sites.
  3. Select Edit on the site you want to customize.
  4. Edit the internal URL.
  5. Select Save changes.

When enabled, the Admin area for Geo shows replication details for each site directly from the primary site's UI, and through the Geo secondary proxy, if enabled.

WARNING: We recommend using an HTTPS connection while configuring the Geo sites. To avoid breaking communication between primary and secondary sites when using HTTPS, customize your Internal URL to point to a load balancer with TLS terminated at the load balancer.

Multiple secondary sites behind a load balancer

Secondary sites can use identical external URLs if a unique name is set for each Geo site. The gitlab.rb setting gitlab_rails['geo_node_name'] must:

  • Be set for each GitLab instance that runs puma, sidekiq, or geo_logcursor.
  • Match a Geo site name.

The load balancer must use sticky sessions to avoid authentication failures and cross-site request errors.