Network access

There are two separate lists: what your own network has to allow, and what your pipelines are allowed to reach.

What your IT department has to allow

Give this list to your IT department. All entries are HTTPS on port 443 unless stated otherwise.

Host Port Purpose
code.vector.cloud 443 GitLab web interface and Git over HTTPS
code.vector.cloud 22 Git over SSH
*.code.vector.cloud 443 Container registry, GitLab Pages and other platform subdomains
auth.vector.com 443 Sign-in
gitlab-prod-vector-release-artifacts.s3.eu-central-1.amazonaws.com 443 Release artifacts
gitlab-prod-vector-release-packages.s3.eu-central-1.amazonaws.com 443 Release packages

The wildcard covers registry.code.vector.cloud (container registry) and pages.code.vector.cloud (GitLab Pages, including this site).

If your IT cannot allow wildcards

Pages traffic is served through a load balancer with static IP addresses reserved for it, so allow-listing by address is possible. Ask your Vector contact for the current addresses. Do not resolve the hostnames yourself: a resolved address can change, but the reserved addresses are stable.

Checking it works

From a computer on your corporate network:

curl -I https://code.vector.cloud               # web and Git over HTTPS
ssh -T git@code.vector.cloud                    # expect "Welcome to GitLab", not a shell
curl -I https://registry.code.vector.cloud/v2/  # container registry

A timeout usually means a firewall restriction. A TLS error usually means an intercepting proxy. Tell your IT department the connection is being inspected, and confirm with your Vector contact whether that is expected for your setup.

What your pipelines can reach

The list above covers your computers. Jobs run on your runners, in a network operated by Vector, where outbound traffic is deny-all with an allow list. A common set of destinations is permitted; anything else has to be added.

Report external hosts during onboarding

If your jobs pull from your own artifact server, an internal package index, or a vendor-specific host, say so during onboarding. Vector adds each firewall rule manually, and this takes several days.

Send hostnames and IP ranges and ports to your Vector contact, or to support@vector.com.