System requirements

Hardware and port requirements

Before installing Continuous Delivery for PE, ensure that your system meets these requirements.

Hardware requirements

While Continuous Delivery for PE can be installed on a single server, installing on three servers is recommended to provide redundancy in case one server becomes unavailable.

Each server must meet the following minimum requirements:


Memory	Storage	CPUs
16 GB	At least 200 GB for `/opt/replicated` At least 32 GB for `/var/lib/kubelet` At least 40 GB for `/var/lib/containerd` Note: If you first installed the platform admin console prior to 7 December 2020, you can safely continue to allocate 40 GB of storage to `/var/lib/docker` instead.	8 (or equivalent)

Note: If Continuous Delivery for PE is used by more than one team in your organization, allocate additional storage space to /opt/replicated to support the increased load on the system.

Open port requirements

Continuous Delivery for PE uses the following ports:


Port	Protocol	Purpose	Source	Destination
Continuous Delivery for PE ports
443	TCP	Web UI	Browser	Continuous Delivery for PE
8000	TCP	Webhook service	Source control	Continuous Delivery for PE
Platform ports
2379	TCP	High availability (HA) communication	etcd on the Kubernetes host	etcd on the Kubernetes host
2380	TCP	High availability (HA) communication	etcd on the Kubernetes host	etcd on the Kubernetes host
6443	TCP	Kubernetes API	Admin workstation	Kubernetes host
6783	TCP/UDP	Kubernetes networking - Weave	Kubernetes host	Kubernetes host
6784	UDP	Kubernetes networking - Weave	Kubernetes host	Kubernetes host
8800	TCP	Platform admin console	Admin browser	Kubernetes host
9001 (offline installations only)	TCP	Local registry for offline installations	Job hardware	Kubernetes host
10250	TCP	Kubernetes cluster management	Kubernetes host	Kubernetes host

IP address range requirements

IP address ranges 10.96.0.0/22 and 10.32.0.0/22 must be locally accessible. See Resolve IP address range conflicts for instructions.

Web URL requirements for firewalls

Continuous Delivery for PE interacts with external web URLs for a variety of installation, configuration, upgrade, and deployment tasks. The following list of web URLs are used by Continuous Delivery for PE for internal and outbound network traffic.


Category	URLs
Platform admin console and platform	get.replicated.com registry.replicated.com proxy.replicated.com api.replicated.com k8s.kurl.sh kurl-sh.s3.amazonaws.com replicated.app registry-data.replicated.com
Container registries	gcr.io docker.io index.docker.io registry-1.docker.io auth.docker.io production.cloudflare.docker.com quay.io
Puppet Enterprise	pup.pt forgeapi.puppet.com pm.puppetlabs.com amazonaws.com s3.amazonaws.com rubygems.org
Source control	The URL of your external source control provider, if applicable.

Supported operating systems

Continuous Delivery for PE can be installed on these operating systems.


Operating system	Supported versions
CentOS	7.4, 7.5, 7.6, 7.7, 7.8, 7.9 8.1, 8.2, 8.3
Red Hat Enterprise Linux (RHEL)	7.4, 7.5, 7.6, 7.7, 7.8, 7.9 8.1, 8.2, 8.3
Ubuntu (General availability kernels)	16.04 (on PE versions prior to 2019.8) 18.04 20.04

Antivirus and antimalware considerations

Antivirus and antimalware software can impact Continuous Delivery for PE or prevent it from functioning properly.

To avoid issues, exclude the following directories from antivirus and antimalware tools that scan disk write operations:

/opt/replicated/rook
/var/lib/docker
/var/lib/kubelet

Job hardware requirements

System requirements for your job hardware vary considerably based on the size of your Continuous Delivery for PE installation, the type of jobs you run, and how frequently you run them.

The size of the load placed on a job hardware server determines how robust that server's resources need to be. Determining that load involves a huge number of variables, from the number of jobs that run concurrently to the languages those jobs are written in. As a result, it's nearly impossible to provide one-size-fits-all system requirements for job hardware.

Instead, we've developed a sizing chart based on the estimated number of concurrent jobs a job hardware server is expected to regularly handle. While this chart represents our best estimates and understanding, it's provided only as a starting point. Testing and experience will help you fine-tune your job hardware and determine the optimum resource configuration for your installation's unique circumstances.


Estimated concurrent job load	Memory	Disk storage	CPUs
2 - 4 concurrent spec tests	4 GB	100 GB	2
4 - 8 concurrent spec tests	8 GB	100 GB	4
6 - 12 concurrent spec tests	8 GB	100 GB	6

When setting up your job hardware, keep these facts in mind:

Disk storage requirements are minimal, and don't increase with added load. After a job run is complete, the job's log is passed to the object storage, and all data related to the job run is erased from the job hardware.
You can run more jobs concurrently without increasing CPUs, but the jobs will run more slowly.

Job hardware requirements for Docker-based jobs

To run Docker-based jobs, your job hardware must have a modern version of Docker CE or Docker EE installed. The puppetlabs/docker module is our preferred way to install Docker and keep it up to date.

Job hardware used for Docker-based jobs also requires internet access. If you're working in an air-gapped environment, set up an internal Docker registry by following the Docker documentation.

Supported Puppet Enterprise versions

The following versions of Puppet Enterprise (PE) are supported for use with Continuous Delivery for PE.


PE version
2019.8.x (LTS)
2019.7
2019.5
2019.4
2019.3

For more on PE versions, see Puppet Enterprise support lifecycle.

Supported browsers

The following browsers are supported for use with the Continuous Delivery for PE web UI.


Browser	Supported versions
Google Chrome	Current version as of release
Mozilla Firefox	Current version as of release
Microsoft Edge	Current version as of release
Apple Safari	Current version as of release