Installing GitHub Enterprise Server on Google Cloud Platform
To install GitHub Enterprise Server on Google Cloud Platform, you must deploy onto a supported machine type and use a persistent standard disk or a persistent SSD.
In this article
- Hardware considerations
- Determining the machine type
- Selecting the GitHub Enterprise Server image
- Configuring the firewall
- Allocating a static IP and assigning it to the VM
- Creating the GitHub Enterprise Server instance
- Configuring the instance
- Further reading
You must have a GitHub Enterprise license file. To download an existing license file or request a trial license, visit enterprise.github.com. For more information, see "Managing your GitHub Enterprise Server license."
You must have a Google Cloud Platform account capable of launching Google Compute Engine (GCE) virtual machine (VM) instances. For more information, see the Google Cloud Platform website and the Google Cloud Platform Documentation.
Most actions needed to launch your instance may also be performed using the Google Cloud Platform Console. However, we recommend installing the gcloud compute command-line tool for initial setup. Examples using the gcloud compute command-line tool are included below. For more information, see the "gcloud compute" installation and setup guide in the Google documentation.
GitHub Enterprise Server requires a persistent data disk separate from the root disk. For more information, see "System overview."
We recommend different hardware configurations based on the number of seats used in your GitHub Enterprise Server instance.
|Seats||vCPUs||Memory||Attached storage||Root storage|
|10-500||2||16 GB||100 GB||200 GB|
|500-3000||4||32 GB||250 GB||200 GB|
|3000-5000||8||64 GB||500 GB||200 GB|
|5000-8000||12||72 GB||750 GB||200 GB|
|8000-10000+||16||128 GB||1000 GB||200 GB|
These are minimum recommendations. More resources may be required depending on your usage, such as user activity and selected integrations. For more information, see "Increasing CPU or memory resources."
Note: The root disk can be resized by building a new appliance or using an existing appliance. For more information, see "Increasing storage capacity."
Warning: We recommend you use webhooks to fetch repository changes for continuous integration (CI) or similar systems. Regular automatic checks, or polling, will significantly reduce the scalability of your instance. For more information, see "About webhooks"
Determining the machine type
Before launching your GitHub Enterprise Server instance on Google Cloud Platform, you'll need to determine the machine type that best fits the needs of your organization.
Supported machine types
GitHub Enterprise Server is supported on the following Google Compute Engine (GCE) machine types. For more information, see the Google Cloud Platform machine types article.
Recommended machine types
Based on your seat count, we recommend these machine types.
|10 - 500||n1-standard-4|
|500 - 3000||n1-standard-8|
|3000 - 5000||n1-highmem-8|
|5000 - 8000||n1-highmem-16|
|8000 - 10000+||n1-highmem-32|
Note: You can always scale up your CPU or memory by resizing your instance. However, because resizing your CPU or memory requires downtime for your users, we recommend over-provisioning resources to account for scale.
Selecting the GitHub Enterprise Server image
Using the gcloud compute command-line tool, list the public GitHub Enterprise Server images:
$ gcloud compute images list --project github-enterprise-public --no-standard-images
Take note of the image name for the latest GCE image of GitHub Enterprise Server.
Configuring the firewall
GCE virtual machines are created as a member of a network, which has a firewall. For the network associated with the GitHub Enterprise Server VM, you'll need to configure the firewall to allow the required ports listed in the table below. For more information about firewall rules on Google Cloud Platform, see the Google guide "Firewall Rules Overview."
Using the gcloud compute command-line tool, create the network. For more information, see "gcloud compute networks create" in the Google documentation.
$ gcloud compute networks create NETWORK-NAME --subnet-mode auto
Create a firewall rule for each of the ports in the table below. For more information, see "gcloud compute firewall-rules" in the Google documentation.
$ gcloud compute firewall-rules create RULE-NAME \ --network NETWORK-NAME \ --allow tcp:22,tcp:25,tcp:80,tcp:122,udp:161,tcp:443,udp:1194,tcp:8080,tcp:8443,tcp:9418,icmp
This table identifies the required ports and what each port is used for.
Port Service Description 22 SSH Git over SSH access. Clone, fetch, and push operations to public/private repositories supported. 25 SMTP SMTP with encryption (STARTTLS) support. 80 HTTP Web application access. All requests are redirected to the HTTPS port when SSL is enabled. 122 SSH Instance shell access. The default SSH port (22) is dedicated to application git+ssh network traffic. 161/UDP SNMP Required for network monitoring protocol operation. 443 HTTPS Web application and Git over HTTPS access. 1194/UDP VPN Secure replication network tunnel in high availability configuration. 8080 HTTP Plain-text web based Management Console. Not required unless SSL is disabled manually. 8443 HTTPS Secure web based Management Console. Required for basic installation and configuration. 9418 Git Simple Git protocol port. Clone and fetch operations to public repositories only. Unencrypted network communication.
Allocating a static IP and assigning it to the VM
If this is a production appliance, we strongly recommend reserving a static external IP address and assigning it to the GitHub Enterprise Server VM. Otherwise, the public IP address of the VM will not be retained after restarts. For more information, see the Google guide "Reserving a Static External IP Address."
In production High Availability configurations, both primary and replica appliances should be assigned separate static IP addresses.
Creating the GitHub Enterprise Server instance
To create the GitHub Enterprise Server instance, you'll need to create a GCE instance with your GitHub Enterprise Server image and attach an additional storage volume for your instance data. For more information, see "Hardware considerations."
Using the gcloud compute command-line tool, create a data disk to use as an attached storage volume for your instance data, and configure the size based on your seat count. For more information, see "gcloud compute disks create" in the Google documentation.
$ gcloud compute disks create DATA-DISK-NAME --size DATA-DISK-SIZE --type DATA-DISK-TYPE --zone ZONE
Then create an instance using the name of the GitHub Enterprise Server image you selected, and attach the data disk. For more information, see "gcloud compute instances create" in the Google documentation.
$ gcloud compute instances create INSTANCE-NAME \ --machine-type n1-standard-8 \ --image GITHUB-ENTERPRISE-IMAGE-NAME \ --disk name=DATA-DISK-NAME \ --metadata serial-port-enable=1 \ --zone ZONE \ --network NETWORK-NAME \ --image-project github-enterprise-public
Configuring the instance
Copy the virtual machine's public DNS name, and paste it into a web browser.
At the prompt, upload your license file and set a management console password. For more information, see "Managing your GitHub Enterprise Server license."
In the Management Console, configure and save your desired settings. For more information, see "Configuring the GitHub Enterprise Server appliance."
The instance will restart automatically.
Click Visit your instance.