Setting up security groups and access restrictions to a Managed Service for GitLab instance
Security group rules determine the following:
- IPs that can access the instance, including web access.
- Protocol to work with Git repositories in the GitLab instance: SSH or HTTPS.
- Certificate for HTTPS: Let's Encrypt
(default) or your own certificate . - Whether or not access to GitLab Container Registry
is provided.
Warning
The security group's setup determines the Managed Service for GitLab instance performance and availability.
To configure a security group for a Managed Service for GitLab instance:
- Add rules for incoming and outgoing traffic to the existing security group or create a new group with such rules.
- Apply the security group to the GitLab instance when creating or modifying it.
If you do not bind a separate security group to an instance, the group created by default in the instance's network will apply to it. The rules of this security group added for other services affect access to the GitLab instance.
If you have issues with setting up a security group, contact support
Rules for incoming traffic
|
Rule purpose |
Rule settings |
|
To access Git repositories over SSH. |
|
|
To access Git repositories over HTTPS. |
|
|
To enable Let’s Encrypt certificate. This certificate is used by default |
|
|
For creating instance backups. |
|
|
For health checks by a network load balancer. |
|
|
To connect to GitLab Container Registry. |
|
Rules for outgoing traffic
Managed Service for GitLab relies on third-party integrations to provide its services. If you limit the outgoing traffic in the instance's security group, the instance may work incorrectly. To avoid this, add the following rules to the security group:
|
Rule purpose |
Rule settings |
|
To enable Let’s Encrypt certificate. |
|
|
For creating instance backups. |
|
|
For requests to the metadata service when updating an instance. |
|
|
For requests to the DNS service. |
|
|
For requests to NTP servers to support two-factor authentication. |
|
|
For access to workers managed by a runner created via the management console. |
|