Installing a user-provisioned bare metal cluster on a restricted network

In OKD 4.10, you can install a cluster on bare metal infrastructure that you provision in a restricted network.

While you might be able to follow this procedure to deploy a cluster on virtualized or cloud environments, you must be aware of additional considerations for non-bare metal platforms. Review the information in the guidelines for deploying OKD on non-tested platforms before you attempt to install an OKD cluster in such an environment.

Prerequisites

About installations in restricted networks

In OKD 4.10, you can perform an installation that does not require an active connection to the internet to obtain software components. Restricted network installations can be completed using installer-provisioned infrastructure or user-provisioned infrastructure, depending on the cloud platform to which you are installing the cluster.

If you choose to perform a restricted network installation on a cloud platform, you still require access to its cloud APIs. Some cloud functions, like Amazon Web Service’s IAM service, require internet access, so you might still require internet access. Depending on your network, you might require less internet access for an installation on bare metal hardware or on VMware vSphere.

To complete a restricted network installation, you must create a registry that mirrors the contents of the OKD registry and contains the installation media. You can create this registry on a mirror host, which can access both the internet and your closed network, or by using other methods that meet your restrictions.

Because of the complexity of the configuration for user-provisioned installations, consider completing a standard user-provisioned infrastructure installation before you attempt a restricted network installation using user-provisioned infrastructure. Completing this test installation might make it easier to isolate and troubleshoot any issues that might arise during your installation in a restricted network.

Additional limits

Clusters in restricted networks have the following additional limitations and restrictions:

  • The ClusterVersion status includes an Unable to retrieve available updates error.

  • By default, you cannot use the contents of the Developer Catalog because you cannot access the required image stream tags.

Requirements for a cluster with user-provisioned infrastructure

For a cluster that contains user-provisioned infrastructure, you must deploy all of the required machines.

This section describes the requirements for deploying OKD on user-provisioned infrastructure.

Required machines for cluster installation

The smallest OKD clusters require the following hosts:

Table 1. Minimum required hosts
HostsDescription

One temporary bootstrap machine

The cluster requires the bootstrap machine to deploy the OKD cluster on the three control plane machines. You can remove the bootstrap machine after you install the cluster.

Three control plane machines

The control plane machines run the Kubernetes and OKD services that form the control plane.

At least two compute machines, which are also known as worker machines.

The workloads requested by OKD users run on the compute machines.

As an exception, you can run zero compute machines in a bare metal cluster that consists of three control plane machines only. This provides smaller, more resource efficient clusters for cluster administrators and developers to use for testing, development, and production. Running one compute machine is not supported.

To maintain high availability of your cluster, use separate physical hosts for these cluster machines.

The bootstrap and control plane machines must use Fedora CoreOS (FCOS) as the operating system. However, the compute machines can choose between Fedora CoreOS (FCOS), Fedora 8.4, or Fedora 8.5.

See Red Hat Enterprise Linux technology capabilities and limits.

Minimum resource requirements for cluster installation

Each cluster machine must meet the following minimum requirements:

MachineOperating SystemCPU [1]RAMStorageIOPS [2]

Bootstrap

FCOS

4

16 GB

100 GB

300

Control plane

FCOS

4

16 GB

100 GB

300

Compute

FCOS

2

8 GB

100 GB

300

  1. One CPU is equivalent to one physical core when simultaneous multithreading (SMT), or hyperthreading, is not enabled. When enabled, use the following formula to calculate the corresponding ratio: (threads per core × cores) × sockets = CPUs.

  2. OKD and Kubernetes are sensitive to disk performance, and faster storage is recommended, particularly for etcd on the control plane nodes which require a 10 ms p99 fsync duration. Note that on many cloud platforms, storage size and IOPS scale together, so you might need to over-allocate storage volume to obtain sufficient performance.

  3. As with all user-provisioned installations, if you choose to use Fedora compute machines in your cluster, you take responsibility for all operating system life cycle management and maintenance, including performing system updates, applying patches, and completing all other required tasks. Use of Fedora 7 compute machines is deprecated and planned for removal in a future release of OKD 4.

Certificate signing requests management

Because your cluster has limited access to automatic machine management when you use infrastructure that you provision, you must provide a mechanism for approving cluster certificate signing requests (CSRs) after installation. The kube-controller-manager only approves the kubelet client CSRs. The machine-approver cannot guarantee the validity of a serving certificate that is requested by using kubelet credentials because it cannot confirm that the correct machine issued the request. You must determine and implement a method of verifying the validity of the kubelet serving certificate requests and approving them.

Additional resources

Networking requirements for user-provisioned infrastructure

All the Fedora CoreOS (FCOS) machines require networking to be configured in initramfs during boot to fetch their Ignition config files.

During the initial boot, the machines require an IP address configuration that is set either through a DHCP server or statically by providing the required boot options. After a network connection is established, the machines download their Ignition config files from an HTTP or HTTPS server. The Ignition config files are then used to set the exact state of each machine. The Machine Config Operator completes more changes to the machines, such as the application of new certificates or keys, after installation.

It is recommended to use a DHCP server for long-term management of the cluster machines. Ensure that the DHCP server is configured to provide persistent IP addresses, DNS server information, and hostnames to the cluster machines.

If a DHCP service is not available for your user-provisioned infrastructure, you can instead provide the IP networking configuration and the address of the DNS server to the nodes at FCOS install time. These can be passed as boot arguments if you are installing from an ISO image. See the Installing FCOS and starting the OKD bootstrap process section for more information about static IP provisioning and advanced networking options.

The Kubernetes API server must be able to resolve the node names of the cluster machines. If the API servers and worker nodes are in different zones, you can configure a default DNS search zone to allow the API server to resolve the node names. Another supported approach is to always refer to hosts by their fully-qualified domain names in both the node objects and all DNS requests.

Setting the cluster node hostnames through DHCP

On Fedora CoreOS (FCOS) machines, the hostname is set through NetworkManager. By default, the machines obtain their hostname through DHCP. If the hostname is not provided by DHCP, it is obtained through a reverse DNS lookup. Reverse DNS lookup occurs after the network has been initialized on a node and can take time to resolve. Other system services can start prior to this and detect the hostname as localhost or similar. You can avoid this by using DHCP to provide the hostname for each cluster node.

Additionally, setting the hostnames through DHCP can bypass any manual DNS record name configuration errors in environments that have a DNS split-horizon implementation.

Network connectivity requirements

You must configure the network connectivity between machines to allow OKD cluster components to communicate. Each machine must be able to resolve the hostnames of all other machines in the cluster.

This section provides details about the ports that are required.

Table 2. Ports used for all-machine to all-machine communications
ProtocolPortDescription

ICMP

N/A

Network reachability tests

TCP

1936

Metrics

9000-9999

Host level services, including the node exporter on ports 9100-9101 and the Cluster Version Operator on port 9099.

10250-10259

The default ports that Kubernetes reserves

10256

openshift-sdn

UDP

4789

VXLAN

6081

Geneve

9000-9999

Host level services, including the node exporter on ports 9100-9101.

TCP/UDP

30000-32767

Kubernetes node port

Table 3. Ports used for all-machine to control plane communications
ProtocolPortDescription

TCP

6443

Kubernetes API

Table 4. Ports used for control plane machine to control plane machine communications
ProtocolPortDescription

TCP

2379-2380

etcd server and peer ports

NTP configuration for user-provisioned infrastructure

OKD clusters are configured to use a public Network Time Protocol (NTP) server by default. If you want to use a local enterprise NTP server, or if your cluster is being deployed in a disconnected network, you can configure the cluster to use a specific time server. For more information, see the documentation for Configuring chrony time service.

If a DHCP server provides NTP server information, the chrony time service on the Fedora CoreOS (FCOS) machines read the information and can sync the clock with the NTP servers.

Additional resources

User-provisioned DNS requirements

In OKD deployments, DNS name resolution is required for the following components:

  • The Kubernetes API

  • The OKD application wildcard

  • The bootstrap, control plane, and compute machines

Reverse DNS resolution is also required for the Kubernetes API, the bootstrap machine, the control plane machines, and the compute machines.

DNS A/AAAA or CNAME records are used for name resolution and PTR records are used for reverse name resolution. The reverse records are important because Fedora CoreOS (FCOS) uses the reverse records to set the hostnames for all the nodes, unless the hostnames are provided by DHCP. Additionally, the reverse records are used to generate the certificate signing requests (CSR) that OKD needs to operate.

It is recommended to use a DHCP server to provide the hostnames to each cluster node. See the DHCP recommendations for user-provisioned infrastructure section for more information.

The following DNS records are required for a user-provisioned OKD cluster and they must be in place before installation. In each record, <cluster_name> is the cluster name and <base_domain> is the base domain that you specify in the install-config.yaml file. A complete DNS record takes the form: <component>.<cluster_name>.<base_domain>..

Table 5. Required DNS records
ComponentRecordDescription

Kubernetes API

api.<cluster_name>.<base_domain>.

A DNS A/AAAA or CNAME record, and a DNS PTR record, to identify the API load balancer. These records must be resolvable by both clients external to the cluster and from all the nodes within the cluster.

api-int.<cluster_name>.<base_domain>.

A DNS A/AAAA or CNAME record, and a DNS PTR record, to internally identify the API load balancer. These records must be resolvable from all the nodes within the cluster.

The API server must be able to resolve the worker nodes by the hostnames that are recorded in Kubernetes. If the API server cannot resolve the node names, then proxied API calls can fail, and you cannot retrieve logs from pods.

Routes

*.apps.<cluster_name>.<base_domain>.

A wildcard DNS A/AAAA or CNAME record that refers to the application ingress load balancer. The application ingress load balancer targets the machines that run the Ingress Controller pods. The Ingress Controller pods run on the compute machines by default. These records must be resolvable by both clients external to the cluster and from all the nodes within the cluster.

For example, console-openshift-console.apps.<cluster_name>.<base_domain> is used as a wildcard route to the OKD console.

Bootstrap machine

bootstrap.<cluster_name>.<base_domain>.

A DNS A/AAAA or CNAME record, and a DNS PTR record, to identify the bootstrap machine. These records must be resolvable by the nodes within the cluster.

Control plane machines

<master><n>.<cluster_name>.<base_domain>.

DNS A/AAAA or CNAME records and DNS PTR records to identify each machine for the control plane nodes. These records must be resolvable by the nodes within the cluster.

Compute machines

<worker><n>.<cluster_name>.<base_domain>.

DNS A/AAAA or CNAME records and DNS PTR records to identify each machine for the worker nodes. These records must be resolvable by the nodes within the cluster.

In OKD 4.4 and later, you do not need to specify etcd host and SRV records in your DNS configuration.

You can use the dig command to verify name and reverse name resolution. See the section on Validating DNS resolution for user-provisioned infrastructure for detailed validation steps.

Example DNS configuration for user-provisioned clusters

This section provides A and PTR record configuration samples that meet the DNS requirements for deploying OKD on user-provisioned infrastructure. The samples are not meant to provide advice for choosing one DNS solution over another.

In the examples, the cluster name is ocp4 and the base domain is example.com.

Example DNS A record configuration for a user-provisioned cluster

The following example is a BIND zone file that shows sample A records for name resolution in a user-provisioned cluster.

Sample DNS zone database

  1. $TTL 1W
  2. @ IN SOA ns1.example.com. root (
  3. 2019070700 ; serial
  4. 3H ; refresh (3 hours)
  5. 30M ; retry (30 minutes)
  6. 2W ; expiry (2 weeks)
  7. 1W ) ; minimum (1 week)
  8. IN NS ns1.example.com.
  9. IN MX 10 smtp.example.com.
  10. ;
  11. ;
  12. ns1.example.com. IN A 192.168.1.5
  13. smtp.example.com. IN A 192.168.1.5
  14. ;
  15. helper.example.com. IN A 192.168.1.5
  16. helper.ocp4.example.com. IN A 192.168.1.5
  17. ;
  18. api.ocp4.example.com. IN A 192.168.1.5 (1)
  19. api-int.ocp4.example.com. IN A 192.168.1.5 (2)
  20. ;
  21. *.apps.ocp4.example.com. IN A 192.168.1.5 (3)
  22. ;
  23. bootstrap.ocp4.example.com. IN A 192.168.1.96 (4)
  24. ;
  25. master0.ocp4.example.com. IN A 192.168.1.97 (5)
  26. master1.ocp4.example.com. IN A 192.168.1.98 (5)
  27. master2.ocp4.example.com. IN A 192.168.1.99 (5)
  28. ;
  29. worker0.ocp4.example.com. IN A 192.168.1.11 (6)
  30. worker1.ocp4.example.com. IN A 192.168.1.7 (6)
  31. ;
  32. ;EOF
1Provides name resolution for the Kubernetes API. The record refers to the IP address of the API load balancer.
2Provides name resolution for the Kubernetes API. The record refers to the IP address of the API load balancer and is used for internal cluster communications.
3Provides name resolution for the wildcard routes. The record refers to the IP address of the application ingress load balancer. The application ingress load balancer targets the machines that run the Ingress Controller pods. The Ingress Controller pods run on the compute machines by default.

In the example, the same load balancer is used for the Kubernetes API and application ingress traffic. In production scenarios, you can deploy the API and application ingress load balancers separately so that you can scale the load balancer infrastructure for each in isolation.

4Provides name resolution for the bootstrap machine.
5Provides name resolution for the control plane machines.
6Provides name resolution for the compute machines.

Example DNS PTR record configuration for a user-provisioned cluster

The following example BIND zone file shows sample PTR records for reverse name resolution in a user-provisioned cluster.

Sample DNS zone database for reverse records

  1. $TTL 1W
  2. @ IN SOA ns1.example.com. root (
  3. 2019070700 ; serial
  4. 3H ; refresh (3 hours)
  5. 30M ; retry (30 minutes)
  6. 2W ; expiry (2 weeks)
  7. 1W ) ; minimum (1 week)
  8. IN NS ns1.example.com.
  9. ;
  10. 5.1.168.192.in-addr.arpa. IN PTR api.ocp4.example.com. (1)
  11. 5.1.168.192.in-addr.arpa. IN PTR api-int.ocp4.example.com. (2)
  12. ;
  13. 96.1.168.192.in-addr.arpa. IN PTR bootstrap.ocp4.example.com. (3)
  14. ;
  15. 97.1.168.192.in-addr.arpa. IN PTR master0.ocp4.example.com. (4)
  16. 98.1.168.192.in-addr.arpa. IN PTR master1.ocp4.example.com. (4)
  17. 99.1.168.192.in-addr.arpa. IN PTR master2.ocp4.example.com. (4)
  18. ;
  19. 11.1.168.192.in-addr.arpa. IN PTR worker0.ocp4.example.com. (5)
  20. 7.1.168.192.in-addr.arpa. IN PTR worker1.ocp4.example.com. (5)
  21. ;
  22. ;EOF
1Provides reverse DNS resolution for the Kubernetes API. The PTR record refers to the record name of the API load balancer.
2Provides reverse DNS resolution for the Kubernetes API. The PTR record refers to the record name of the API load balancer and is used for internal cluster communications.
3Provides reverse DNS resolution for the bootstrap machine.
4Provides reverse DNS resolution for the control plane machines.
5Provides reverse DNS resolution for the compute machines.

A PTR record is not required for the OKD application wildcard.

Additional resources

Load balancing requirements for user-provisioned infrastructure

Before you install OKD, you must provision the API and application ingress load balancing infrastructure. In production scenarios, you can deploy the API and application ingress load balancers separately so that you can scale the load balancer infrastructure for each in isolation.

The load balancing infrastructure must meet the following requirements:

  1. API load balancer: Provides a common endpoint for users, both human and machine, to interact with and configure the platform. Configure the following conditions:

    • Layer 4 load balancing only. This can be referred to as Raw TCP, SSL Passthrough, or SSL Bridge mode. If you use SSL Bridge mode, you must enable Server Name Indication (SNI) for the API routes.

    • A stateless load balancing algorithm. The options vary based on the load balancer implementation.

    Session persistence is not required for the API load balancer to function properly.

    Configure the following ports on both the front and back of the load balancers:

    Table 6. API load balancer
    PortBack-end machines (pool members)InternalExternalDescription

    6443

    Bootstrap and control plane. You remove the bootstrap machine from the load balancer after the bootstrap machine initializes the cluster control plane. You must configure the /readyz endpoint for the API server health check probe.

    X

    X

    Kubernetes API server

    22623

    Bootstrap and control plane. You remove the bootstrap machine from the load balancer after the bootstrap machine initializes the cluster control plane.

    X

    Machine config server

    The load balancer must be configured to take a maximum of 30 seconds from the time the API server turns off the /readyz endpoint to the removal of the API server instance from the pool. Within the time frame after /readyz returns an error or becomes healthy, the endpoint must have been removed or added. Probing every 5 or 10 seconds, with two successful requests to become healthy and three to become unhealthy, are well-tested values.

  2. Application ingress load balancer: Provides an ingress point for application traffic flowing in from outside the cluster. Configure the following conditions:

    • Layer 4 load balancing only. This can be referred to as Raw TCP, SSL Passthrough, or SSL Bridge mode. If you use SSL Bridge mode, you must enable Server Name Indication (SNI) for the ingress routes.

    • A connection-based or session-based persistence is recommended, based on the options available and types of applications that will be hosted on the platform.

    If the true IP address of the client can be seen by the application ingress load balancer, enabling source IP-based session persistence can improve performance for applications that use end-to-end TLS encryption.

    Configure the following ports on both the front and back of the load balancers:

    Table 7. Application ingress load balancer
    PortBack-end machines (pool members)InternalExternalDescription

    443

    The machines that run the Ingress Controller pods, compute, or worker, by default.

    X

    X

    HTTPS traffic

    80

    The machines that run the Ingress Controller pods, compute, or worker, by default.

    X

    X

    HTTP traffic

    1936

    The worker nodes that run the Ingress Controller pods, by default. You must configure the /healthz/ready endpoint for the ingress health check probe.

    X

    X

    HTTP traffic

If you are deploying a three-node cluster with zero compute nodes, the Ingress Controller pods run on the control plane nodes. In three-node cluster deployments, you must configure your application ingress load balancer to route HTTP and HTTPS traffic to the control plane nodes.

A working configuration for the Ingress router is required for an OKD cluster. You must configure the Ingress router after the control plane initializes.

Example load balancer configuration for user-provisioned clusters

This section provides an example API and application ingress load balancer configuration that meets the load balancing requirements for user-provisioned clusters. The sample is an /etc/haproxy/haproxy.cfg configuration for an HAProxy load balancer. The example is not meant to provide advice for choosing one load balancing solution over another.

In the example, the same load balancer is used for the Kubernetes API and application ingress traffic. In production scenarios you can deploy the API and application ingress load balancers separately so that you can scale the load balancer infrastructure for each in isolation.

Sample API and application ingress load balancer configuration

  1. global
  2. log 127.0.0.1 local2
  3. pidfile /var/run/haproxy.pid
  4. maxconn 4000
  5. daemon
  6. defaults
  7. mode http
  8. log global
  9. option dontlognull
  10. option http-server-close
  11. option redispatch
  12. retries 3
  13. timeout http-request 10s
  14. timeout queue 1m
  15. timeout connect 10s
  16. timeout client 1m
  17. timeout server 1m
  18. timeout http-keep-alive 10s
  19. timeout check 10s
  20. maxconn 3000
  21. frontend stats
  22. bind *:1936
  23. mode http
  24. log global
  25. maxconn 10
  26. stats enable
  27. stats hide-version
  28. stats refresh 30s
  29. stats show-node
  30. stats show-desc Stats for ocp4 cluster (1)
  31. stats auth admin:ocp4
  32. stats uri /stats
  33. listen api-server-6443 (2)
  34. bind *:6443
  35. mode tcp
  36. server bootstrap bootstrap.ocp4.example.com:6443 check inter 1s backup (3)
  37. server master0 master0.ocp4.example.com:6443 check inter 1s
  38. server master1 master1.ocp4.example.com:6443 check inter 1s
  39. server master2 master2.ocp4.example.com:6443 check inter 1s
  40. listen machine-config-server-22623 (4)
  41. bind *:22623
  42. mode tcp
  43. server bootstrap bootstrap.ocp4.example.com:22623 check inter 1s backup (3)
  44. server master0 master0.ocp4.example.com:22623 check inter 1s
  45. server master1 master1.ocp4.example.com:22623 check inter 1s
  46. server master2 master2.ocp4.example.com:22623 check inter 1s
  47. listen ingress-router-443 (5)
  48. bind *:443
  49. mode tcp
  50. balance source
  51. server worker0 worker0.ocp4.example.com:443 check inter 1s
  52. server worker1 worker1.ocp4.example.com:443 check inter 1s
  53. listen ingress-router-80 (6)
  54. bind *:80
  55. mode tcp
  56. balance source
  57. server worker0 worker0.ocp4.example.com:80 check inter 1s
  58. server worker1 worker1.ocp4.example.com:80 check inter 1s
1In the example, the cluster name is ocp4.
2Port 6443 handles the Kubernetes API traffic and points to the control plane machines.
3The bootstrap entries must be in place before the OKD cluster installation and they must be removed after the bootstrap process is complete.
4Port 22623 handles the machine config server traffic and points to the control plane machines.
5Port 443 handles the HTTPS traffic and points to the machines that run the Ingress Controller pods. The Ingress Controller pods run on the compute machines by default.
6Port 80 handles the HTTP traffic and points to the machines that run the Ingress Controller pods. The Ingress Controller pods run on the compute machines by default.

If you are deploying a three-node cluster with zero compute nodes, the Ingress Controller pods run on the control plane nodes. In three-node cluster deployments, you must configure your application ingress load balancer to route HTTP and HTTPS traffic to the control plane nodes.

If you are using HAProxy as a load balancer, you can check that the haproxy process is listening on ports 6443, 22623, 443, and 80 by running netstat -nltupe on the HAProxy node.

If you are using HAProxy as a load balancer and SELinux is set to enforcing, you must ensure that the HAProxy service can bind to the configured TCP port by running setsebool -P haproxy_connect_any=1.

Preparing the user-provisioned infrastructure

Before you install OKD on user-provisioned infrastructure, you must prepare the underlying infrastructure.

This section provides details about the high-level steps required to set up your cluster infrastructure in preparation for an OKD installation. This includes configuring IP networking and network connectivity for your cluster nodes, enabling the required ports through your firewall, and setting up the required DNS and load balancing infrastructure.

After preparation, your cluster infrastructure must meet the requirements outlined in the Requirements for a cluster with user-provisioned infrastructure section.

Prerequisites

  • You have reviewed the OKD 4.x Tested Integrations page.

  • You have reviewed the infrastructure requirements detailed in the Requirements for a cluster with user-provisioned infrastructure section.

Procedure

  1. If you are using DHCP to provide the IP networking configuration to your cluster nodes, configure your DHCP service.

    1. Add persistent IP addresses for the nodes to your DHCP server configuration. In your configuration, match the MAC address of the relevant network interface to the intended IP address for each node.

    2. When you use DHCP to configure IP addressing for the cluster machines, the machines also obtain the DNS server information through DHCP. Define the persistent DNS server address that is used by the cluster nodes through your DHCP server configuration.

      If you are not using a DHCP service, you must provide the IP networking configuration and the address of the DNS server to the nodes at FCOS install time. These can be passed as boot arguments if you are installing from an ISO image. See the Installing FCOS and starting the OKD bootstrap process section for more information about static IP provisioning and advanced networking options.

    3. Define the hostnames of your cluster nodes in your DHCP server configuration. See the Setting the cluster node hostnames through DHCP section for details about hostname considerations.

      If you are not using a DHCP service, the cluster nodes obtain their hostname through a reverse DNS lookup.

  2. Ensure that your network infrastructure provides the required network connectivity between the cluster components. See the Networking requirements for user-provisioned infrastructure section for details about the requirements.

  3. Configure your firewall to enable the ports required for the OKD cluster components to communicate. See Networking requirements for user-provisioned infrastructure section for details about the ports that are required.

  4. Setup the required DNS infrastructure for your cluster.

    1. Configure DNS name resolution for the Kubernetes API, the application wildcard, the bootstrap machine, the control plane machines, and the compute machines.

    2. Configure reverse DNS resolution for the Kubernetes API, the bootstrap machine, the control plane machines, and the compute machines.

      See the User-provisioned DNS requirements section for more information about the OKD DNS requirements.

  5. Validate your DNS configuration.

    1. From your installation node, run DNS lookups against the record names of the Kubernetes API, the wildcard routes, and the cluster nodes. Validate that the IP addresses in the responses correspond to the correct components.

    2. From your installation node, run reverse DNS lookups against the IP addresses of the load balancer and the cluster nodes. Validate that the record names in the responses correspond to the correct components.

      See the Validating DNS resolution for user-provisioned infrastructure section for detailed DNS validation steps.

  6. Provision the required API and application ingress load balancing infrastructure. See the Load balancing requirements for user-provisioned infrastructure section for more information about the requirements.

Some load balancing solutions require the DNS name resolution for the cluster nodes to be in place before the load balancing is initialized.

Additional resources

Validating DNS resolution for user-provisioned infrastructure

You can validate your DNS configuration before installing OKD on user-provisioned infrastructure.

The validation steps detailed in this section must succeed before you install your cluster.

Prerequisites

  • You have configured the required DNS records for your user-provisioned infrastructure.

Procedure

  1. From your installation node, run DNS lookups against the record names of the Kubernetes API, the wildcard routes, and the cluster nodes. Validate that the IP addresses contained in the responses correspond to the correct components.

    1. Perform a lookup against the Kubernetes API record name. Check that the result points to the IP address of the API load balancer:

      1. $ dig +noall +answer @<nameserver_ip> api.<cluster_name>.<base_domain> (1)
      1Replace <nameserver_ip> with the IP address of the nameserver, <cluster_name> with your cluster name, and <base_domain> with your base domain name.

      Example output

      1. api.ocp4.example.com. 0 IN A 192.168.1.5
    2. Perform a lookup against the Kubernetes internal API record name. Check that the result points to the IP address of the API load balancer:

      1. $ dig +noall +answer @<nameserver_ip> api-int.<cluster_name>.<base_domain>

      Example output

      1. api-int.ocp4.example.com. 0 IN A 192.168.1.5
    3. Test an example *.apps.<cluster_name>.<base_domain> DNS wildcard lookup. All of the application wildcard lookups must resolve to the IP address of the application ingress load balancer:

      1. $ dig +noall +answer @<nameserver_ip> random.apps.<cluster_name>.<base_domain>

      Example output

      1. random.apps.ocp4.example.com. 0 IN A 192.168.1.5

      In the example outputs, the same load balancer is used for the Kubernetes API and application ingress traffic. In production scenarios, you can deploy the API and application ingress load balancers separately so that you can scale the load balancer infrastructure for each in isolation.

      You can replace random with another wildcard value. For example, you can query the route to the OKD console:

      1. $ dig +noall +answer @<nameserver_ip> console-openshift-console.apps.<cluster_name>.<base_domain>

      Example output

      1. console-openshift-console.apps.ocp4.example.com. 0 IN A 192.168.1.5
    4. Run a lookup against the bootstrap DNS record name. Check that the result points to the IP address of the bootstrap node:

      1. $ dig +noall +answer @<nameserver_ip> bootstrap.<cluster_name>.<base_domain>

      Example output

      1. bootstrap.ocp4.example.com. 0 IN A 192.168.1.96
    5. Use this method to perform lookups against the DNS record names for the control plane and compute nodes. Check that the results correspond to the IP addresses of each node.

  2. From your installation node, run reverse DNS lookups against the IP addresses of the load balancer and the cluster nodes. Validate that the record names contained in the responses correspond to the correct components.

    1. Perform a reverse lookup against the IP address of the API load balancer. Check that the response includes the record names for the Kubernetes API and the Kubernetes internal API:

      1. $ dig +noall +answer @<nameserver_ip> -x 192.168.1.5

      Example output

      1. 5.1.168.192.in-addr.arpa. 0 IN PTR api-int.ocp4.example.com. (1)
      2. 5.1.168.192.in-addr.arpa. 0 IN PTR api.ocp4.example.com. (2)
      1Provides the record name for the Kubernetes internal API.
      2Provides the record name for the Kubernetes API.

      A PTR record is not required for the OKD application wildcard. No validation step is needed for reverse DNS resolution against the IP address of the application ingress load balancer.

    2. Perform a reverse lookup against the IP address of the bootstrap node. Check that the result points to the DNS record name of the bootstrap node:

      1. $ dig +noall +answer @<nameserver_ip> -x 192.168.1.96

      Example output

      1. 96.1.168.192.in-addr.arpa. 0 IN PTR bootstrap.ocp4.example.com.
    3. Use this method to perform reverse lookups against the IP addresses for the control plane and compute nodes. Check that the results correspond to the DNS record names of each node.

Additional resources

Generating a key pair for cluster node SSH access

During an OKD installation, you can provide an SSH public key to the installation program. The key is passed to the Fedora CoreOS (FCOS) nodes through their Ignition config files and is used to authenticate SSH access to the nodes. The key is added to the ~/.ssh/authorized_keys list for the core user on each node, which enables password-less authentication.

After the key is passed to the nodes, you can use the key pair to SSH in to the FCOS nodes as the user core. To access the nodes through SSH, the private key identity must be managed by SSH for your local user.

If you want to SSH in to your cluster nodes to perform installation debugging or disaster recovery, you must provide the SSH public key during the installation process. The ./openshift-install gather command also requires the SSH public key to be in place on the cluster nodes.

Do not skip this procedure in production environments, where disaster recovery and debugging is required.

You must use a local key, not one that you configured with platform-specific approaches such as AWS key pairs.

On clusters running Fedora CoreOS (FCOS), the SSH keys specified in the Ignition config files are written to the /home/core/.ssh/authorized_keys.d/core file. However, the Machine Config Operator manages SSH keys in the /home/core/.ssh/authorized_keys file and configures sshd to ignore the /home/core/.ssh/authorized_keys.d/core file. As a result, newly provisioned OKD nodes are not accessible using SSH until the Machine Config Operator reconciles the machine configs with the authorized_keys file. After you can access the nodes using SSH, you can delete the /home/core/.ssh/authorized_keys.d/core file.

Procedure

  1. If you do not have an existing SSH key pair on your local machine to use for authentication onto your cluster nodes, create one. For example, on a computer that uses a Linux operating system, run the following command:

    1. $ ssh-keygen -t ed25519 -N '' -f <path>/<file_name> (1)
    1Specify the path and file name, such as ~/.ssh/id_rsa, of the new SSH key. If you have an existing key pair, ensure your public key is in the your ~/.ssh directory.

    If you plan to install an OKD cluster that uses FIPS Validated / Modules in Process cryptographic libraries on the x86_64 architecture, do not create a key that uses the ed25519 algorithm. Instead, create a key that uses the rsa or ecdsa algorithm.

  2. View the public SSH key:

    1. $ cat <path>/<file_name>.pub

    For example, run the following to view the ~/.ssh/id_rsa.pub public key:

    1. $ cat ~/.ssh/id_rsa.pub
  3. Add the SSH private key identity to the SSH agent for your local user, if it has not already been added. SSH agent management of the key is required for password-less SSH authentication onto your cluster nodes, or if you want to use the ./openshift-install gather command.

    On some distributions, default SSH private key identities such as ~/.ssh/id_rsa and ~/.ssh/id_dsa are managed automatically.

    1. If the ssh-agent process is not already running for your local user, start it as a background task:

      1. $ eval "$(ssh-agent -s)"

      Example output

      1. Agent pid 31874

      If your cluster is in FIPS mode, only use FIPS-compliant algorithms to generate the SSH key. The key must be either RSA or ECDSA.

  4. Add your SSH private key to the ssh-agent:

    1. $ ssh-add <path>/<file_name> (1)
    1Specify the path and file name for your SSH private key, such as ~/.ssh/id_rsa

    Example output

    1. Identity added: /home/<you>/<path>/<file_name> (<computer_name>)

Next steps

  • When you install OKD, provide the SSH public key to the installation program. If you install a cluster on infrastructure that you provision, you must provide the key to the installation program.

Additional resources

Manually creating the installation configuration file

For user-provisioned installations of OKD, you manually generate your installation configuration file.

Prerequisites

  • You have an SSH public key on your local machine to provide to the installation program. The key will be used for SSH authentication onto your cluster nodes for debugging and disaster recovery.

  • You have obtained the OKD installation program and the pull secret for your cluster.

  • Obtain the imageContentSources section from the output of the command to mirror the repository.

  • Obtain the contents of the certificate for your mirror registry.

Procedure

  1. Create an installation directory to store your required installation assets in:

    1. $ mkdir <installation_directory>

    You must create a directory. Some installation assets, like bootstrap X.509 certificates have short expiration intervals, so you must not reuse an installation directory. If you want to reuse individual files from another cluster installation, you can copy them into your directory. However, the file names for the installation assets might change between releases. Use caution when copying installation files from an earlier OKD version.

  2. Customize the sample install-config.yaml file template that is provided and save it in the <installation_directory>.

    You must name this configuration file install-config.yaml.

    • Unless you use a registry that FCOS trusts by default, such as docker.io, you must provide the contents of the certificate for your mirror repository in the additionalTrustBundle section. In most cases, you must provide the certificate for your mirror.

    • You must include the imageContentSources section from the output of the command to mirror the repository.

      For some platform types, you can alternatively run ./openshift-install create install-config —dir <installation_directory> to generate an install-config.yaml file. You can provide details about your cluster configuration at the prompts.

  3. Back up the install-config.yaml file so that you can use it to install multiple clusters.

    The install-config.yaml file is consumed during the next step of the installation process. You must back it up now.

Installation configuration parameters

Before you deploy an OKD cluster, you provide a customized install-config.yaml installation configuration file that describes the details for your environment.

After installation, you cannot modify these parameters in the install-config.yaml file.

Required configuration parameters

Required installation configuration parameters are described in the following table:

Table 8. Required parameters
ParameterDescriptionValues

apiVersion

The API version for the install-config.yaml content. The current version is v1. The installer may also support older API versions.

String

baseDomain

The base domain of your cloud provider. The base domain is used to create routes to your OKD cluster components. The full DNS name for your cluster is a combination of the baseDomain and metadata.name parameter values that uses the <metadata.name>.<baseDomain> format.

A fully-qualified domain or subdomain name, such as example.com.

metadata

Kubernetes resource ObjectMeta, from which only the name parameter is consumed.

Object

metadata.name

The name of the cluster. DNS records for the cluster are all subdomains of {{.metadata.name}}.{{.baseDomain}}.

String of lowercase letters, hyphens (-), and periods (.), such as dev.

platform

The configuration for the specific platform upon which to perform the installation: alibabacloud, aws, baremetal, azure, ibmcloud, openstack, ovirt, vsphere, or {}. For additional information about platform.<platform> parameters, consult the table for your specific platform that follows.

Object

Network configuration parameters

You can customize your installation configuration based on the requirements of your existing network infrastructure. For example, you can expand the IP address block for the cluster network or provide different IP address blocks than the defaults.

Only IPv4 addresses are supported.

Table 9. Network parameters
ParameterDescriptionValues

networking

The configuration for the cluster network.

Object

You cannot modify parameters specified by the networking object after installation.

networking.networkType

The cluster network provider Container Network Interface (CNI) plug-in to install.

Either OpenShiftSDN or OVNKubernetes. The default value is OVNKubernetes.

networking.clusterNetwork

The IP address blocks for pods.

The default value is 10.128.0.0/14 with a host prefix of /23.

If you specify multiple IP address blocks, the blocks must not overlap.

An array of objects. For example:

  1. networking:
  2. clusterNetwork:
  3. - cidr: 10.128.0.0/14
  4. hostPrefix: 23

networking.clusterNetwork.cidr

Required if you use networking.clusterNetwork. An IP address block.

An IPv4 network.

An IP address block in Classless Inter-Domain Routing (CIDR) notation. The prefix length for an IPv4 block is between 0 and 32.

networking.clusterNetwork.hostPrefix

The subnet prefix length to assign to each individual node. For example, if hostPrefix is set to 23 then each node is assigned a /23 subnet out of the given cidr. A hostPrefix value of 23 provides 510 (2^(32 - 23) - 2) pod IP addresses.

A subnet prefix.

The default value is 23.

networking.serviceNetwork

The IP address block for services. The default value is 172.30.0.0/16.

The OpenShift SDN and OVN-Kubernetes network providers support only a single IP address block for the service network.

An array with an IP address block in CIDR format. For example:

  1. networking:
  2. serviceNetwork:
  3. - 172.30.0.0/16

networking.machineNetwork

The IP address blocks for machines.

If you specify multiple IP address blocks, the blocks must not overlap.

An array of objects. For example:

  1. networking:
  2. machineNetwork:
  3. - cidr: 10.0.0.0/16

networking.machineNetwork.cidr

Required if you use networking.machineNetwork. An IP address block. The default value is 10.0.0.0/16 for all platforms other than libvirt. For libvirt, the default value is 192.168.126.0/24.

An IP network block in CIDR notation.

For example, 10.0.0.0/16.

Set the networking.machineNetwork to match the CIDR that the preferred NIC resides in.

Optional configuration parameters

Optional installation configuration parameters are described in the following table:

Table 10. Optional parameters
ParameterDescriptionValues

additionalTrustBundle

A PEM-encoded X.509 certificate bundle that is added to the nodes’ trusted certificate store. This trust bundle may also be used when a proxy has been configured.

String

cgroupsV2

Enables Linux control groups version 2 (cgroups v2) on specific nodes in your cluster. The OKD process for enabling cgroups v2 disables all cgroup version 1 controllers and hierarchies. The OKD cgroups version 2 feature is in Developer Preview and is not supported by Red Hat at this time.

true

compute

The configuration for the machines that comprise the compute nodes.

Array of MachinePool objects.

compute.architecture

Determines the instruction set architecture of the machines in the pool. Currently, clusters with varied architectures are not supported. All pools must specify the same architecture. Valid values are amd64 and arm64.

String

compute.hyperthreading

Whether to enable or disable simultaneous multithreading, or hyperthreading, on compute machines. By default, simultaneous multithreading is enabled to increase the performance of your machines’ cores.

If you disable simultaneous multithreading, ensure that your capacity planning accounts for the dramatically decreased machine performance.

Enabled or Disabled

compute.name

Required if you use compute. The name of the machine pool.

worker

compute.platform

Required if you use compute. Use this parameter to specify the cloud provider to host the worker machines. This parameter value must match the controlPlane.platform parameter value.

alibaba, aws, azure, gcp, ibmcloud, openstack, ovirt, vsphere, or {}

compute.replicas

The number of compute machines, which are also known as worker machines, to provision.

A positive integer greater than or equal to 2. The default value is 3.

controlPlane

The configuration for the machines that comprise the control plane.

Array of MachinePool objects.

controlPlane.architecture

Determines the instruction set architecture of the machines in the pool. Currently, clusters with varied architectures are not supported. All pools must specify the same architecture. Valid values are amd64 and arm64.

String

controlPlane.hyperthreading

Whether to enable or disable simultaneous multithreading, or hyperthreading, on control plane machines. By default, simultaneous multithreading is enabled to increase the performance of your machines’ cores.

If you disable simultaneous multithreading, ensure that your capacity planning accounts for the dramatically decreased machine performance.

Enabled or Disabled

controlPlane.name

Required if you use controlPlane. The name of the machine pool.

master

controlPlane.platform

Required if you use controlPlane. Use this parameter to specify the cloud provider that hosts the control plane machines. This parameter value must match the compute.platform parameter value.

alibaba, aws, azure, gcp, ibmcloud, openstack, ovirt, vsphere, or {}

controlPlane.replicas

The number of control plane machines to provision.

The only supported value is 3, which is the default value.

credentialsMode

The Cloud Credential Operator (CCO) mode. If no mode is specified, the CCO dynamically tries to determine the capabilities of the provided credentials, with a preference for mint mode on the platforms where multiple modes are supported.

Not all CCO modes are supported for all cloud providers. For more information on CCO modes, see the Cloud Credential Operator entry in the Platform Operators reference content.

Mint, Passthrough, Manual, or an empty string (“”).

imageContentSources

Sources and repositories for the release-image content.

Array of objects. Includes a source and, optionally, mirrors, as described in the following rows of this table.

imageContentSources.source

Required if you use imageContentSources. Specify the repository that users refer to, for example, in image pull specifications.

String

imageContentSources.mirrors

Specify one or more repositories that may also contain the same images.

Array of strings

publish

How to publish or expose the user-facing endpoints of your cluster, such as the Kubernetes API, OpenShift routes.

Internal or External. The default value is External.

Setting this field to Internal is not supported on non-cloud platforms and IBM Cloud.

If the value of the field is set to Internal, the cluster will become non-functional. For more information, refer to BZ#1953035.

sshKey

The SSH key or keys to authenticate access your cluster machines.

For production OKD clusters on which you want to perform installation debugging or disaster recovery, specify an SSH key that your ssh-agent process uses.

One or more keys. For example:

  1. sshKey:
  2. <key1>
  3. <key2>
  4. <key3>

Sample install-config.yaml file for bare metal

You can customize the install-config.yaml file to specify more details about your OKD cluster’s platform or modify the values of the required parameters.

  1. apiVersion: v1
  2. baseDomain: example.com (1)
  3. compute: (2)
  4. - hyperthreading: Enabled (3)
  5. name: worker
  6. replicas: 0 (4)
  7. controlPlane: (2)
  8. hyperthreading: Enabled (3)
  9. name: master
  10. replicas: 3 (5)
  11. metadata:
  12. name: test (6)
  13. networking:
  14. clusterNetwork:
  15. - cidr: 10.128.0.0/14 (7)
  16. hostPrefix: 23 (8)
  17. networkType: OVNKubernetes
  18. serviceNetwork: (9)
  19. - 172.30.0.0/16
  20. platform:
  21. none: {} (10)
  22. pullSecret: '{"auths":{"<local_registry>": {"auth": "<credentials>","email": "you@example.com"}}}' (11)
  23. sshKey: 'ssh-ed25519 AAAA...' (12)
  24. additionalTrustBundle: | (13)
  25. -----BEGIN CERTIFICATE-----
  26. ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ
  27. -----END CERTIFICATE-----
  28. imageContentSources: (14)
  29. - mirrors:
  30. - <local_registry>/<local_repository_name>/release
  31. source: quay.io/openshift-release-dev/ocp-release
  32. - mirrors:
  33. - <local_registry>/<local_repository_name>/release
  34. source: quay.io/openshift-release-dev/ocp-v4.0-art-dev
1The base domain of the cluster. All DNS records must be sub-domains of this base and include the cluster name.
2The controlPlane section is a single mapping, but the compute section is a sequence of mappings. To meet the requirements of the different data structures, the first line of the compute section must begin with a hyphen, -, and the first line of the controlPlane section must not. Although both sections currently define a single machine pool, it is possible that future versions of OKD will support defining multiple compute pools during installation. Only one control plane pool is used.
3Specifies whether to enable or disable simultaneous multithreading (SMT), or hyperthreading. By default, SMT is enabled to increase the performance of the cores in your machines. You can disable it by setting the parameter value to Disabled. If you disable SMT, you must disable it in all cluster machines; this includes both control plane and compute machines.

Simultaneous multithreading (SMT) is enabled by default. If SMT is not enabled in your BIOS settings, the hyperthreading parameter has no effect.

If you disable hyperthreading, whether in the BIOS or in the install-config.yaml file, ensure that your capacity planning accounts for the dramatically decreased machine performance.

4You must set this value to 0 when you install OKD on user-provisioned infrastructure. In installer-provisioned installations, the parameter controls the number of compute machines that the cluster creates and manages for you. In user-provisioned installations, you must manually deploy the compute machines before you finish installing the cluster.

If you are installing a three-node cluster, do not deploy any compute machines when you install the Fedora CoreOS (FCOS) machines.

5The number of control plane machines that you add to the cluster. Because the cluster uses these values as the number of etcd endpoints in the cluster, the value must match the number of control plane machines that you deploy.
6The cluster name that you specified in your DNS records.
7A block of IP addresses from which pod IP addresses are allocated. This block must not overlap with existing physical networks. These IP addresses are used for the pod network. If you need to access the pods from an external network, you must configure load balancers and routers to manage the traffic.
8The subnet prefix length to assign to each individual node. For example, if hostPrefix is set to 23, then each node is assigned a /23 subnet out of the given cidr, which allows for 510 (2^(32 - 23) - 2) pod IP addresses. If you are required to provide access to nodes from an external network, configure load balancers and routers to manage the traffic.
9The IP address pool to use for service IP addresses. You can enter only one IP address pool. This block must not overlap with existing physical networks. If you need to access the services from an external network, configure load balancers and routers to manage the traffic.
10You must set the platform to none. You cannot provide additional platform configuration variables for your platform.
11For <local_registry>, specify the registry domain name, and optionally the port, that your mirror registry uses to serve content. For example, registry.example.com or registry.example.com:5000. For <credentials>, specify the base64-encoded user name and password for your mirror registry.
12The SSH public key for the core user in Fedora CoreOS (FCOS).

For production OKD clusters on which you want to perform installation debugging or disaster recovery, specify an SSH key that your ssh-agent process uses.

13Provide the contents of the certificate file that you used for your mirror registry.
14Provide the imageContentSources section from the output of the command to mirror the repository.

Additional resources

Configuring the cluster-wide proxy during installation

Production environments can deny direct access to the internet and instead have an HTTP or HTTPS proxy available. You can configure a new OKD cluster to use a proxy by configuring the proxy settings in the install-config.yaml file.

For bare metal installations, if you do not assign node IP addresses from the range that is specified in the networking.machineNetwork[].cidr field in the install-config.yaml file, you must include them in the proxy.noProxy field.

Prerequisites

  • You have an existing install-config.yaml file.

  • You reviewed the sites that your cluster requires access to and determined whether any of them need to bypass the proxy. By default, all cluster egress traffic is proxied, including calls to hosting cloud provider APIs. You added sites to the Proxy object’s spec.noProxy field to bypass the proxy if necessary.

    The Proxy object status.noProxy field is populated with the values of the networking.machineNetwork[].cidr, networking.clusterNetwork[].cidr, and networking.serviceNetwork[] fields from your installation configuration.

    For installations on Amazon Web Services (AWS), Google Cloud Platform (GCP), Microsoft Azure, and Red Hat OpenStack Platform (RHOSP), the Proxy object status.noProxy field is also populated with the instance metadata endpoint (169.254.169.254).

Procedure

  1. Edit your install-config.yaml file and add the proxy settings. For example:

    1. apiVersion: v1
    2. baseDomain: my.domain.com
    3. proxy:
    4. httpProxy: http://<username>:<pswd>@<ip>:<port> (1)
    5. httpsProxy: https://<username>:<pswd>@<ip>:<port> (2)
    6. noProxy: example.com (3)
    7. additionalTrustBundle: | (4)
    8. -----BEGIN CERTIFICATE-----
    9. <MY_TRUSTED_CA_CERT>
    10. -----END CERTIFICATE-----
    11. ...
    1A proxy URL to use for creating HTTP connections outside the cluster. The URL scheme must be http.
    2A proxy URL to use for creating HTTPS connections outside the cluster.
    3A comma-separated list of destination domain names, IP addresses, or other network CIDRs to exclude from proxying. Preface a domain with . to match subdomains only. For example, .y.com matches x.y.com, but not y.com. Use * to bypass the proxy for all destinations.
    4If provided, the installation program generates a config map that is named user-ca-bundle in the openshift-config namespace that contains one or more additional CA certificates that are required for proxying HTTPS connections. The Cluster Network Operator then creates a trusted-ca-bundle config map that merges these contents with the Fedora CoreOS (FCOS) trust bundle, and this config map is referenced in the trustedCA field of the Proxy object. The additionalTrustBundle field is required unless the proxy’s identity certificate is signed by an authority from the FCOS trust bundle.

    The installation program does not support the proxy readinessEndpoints field.

  2. Save the file and reference it when installing OKD.

The installation program creates a cluster-wide proxy that is named cluster that uses the proxy settings in the provided install-config.yaml file. If no proxy settings are provided, a cluster Proxy object is still created, but it will have a nil spec.

Only the Proxy object named cluster is supported, and no additional proxies can be created.

Configuring a three-node cluster

Optionally, you can deploy zero compute machines in a bare metal cluster that consists of three control plane machines only. This provides smaller, more resource efficient clusters for cluster administrators and developers to use for testing, development, and production.

In three-node OKD environments, the three control plane machines are schedulable, which means that your application workloads are scheduled to run on them.

Prerequisites

  • You have an existing install-config.yaml file.

Procedure

  • Ensure that the number of compute replicas is set to 0 in your install-config.yaml file, as shown in the following compute stanza:

    1. compute:
    2. - name: worker
    3. platform: {}
    4. replicas: 0

    You must set the value of the replicas parameter for the compute machines to 0 when you install OKD on user-provisioned infrastructure, regardless of the number of compute machines you are deploying. In installer-provisioned installations, the parameter controls the number of compute machines that the cluster creates and manages for you. This does not apply to user-provisioned installations, where the compute machines are deployed manually.

For three-node cluster installations, follow these next steps:

  • If you are deploying a three-node cluster with zero compute nodes, the Ingress Controller pods run on the control plane nodes. In three-node cluster deployments, you must configure your application ingress load balancer to route HTTP and HTTPS traffic to the control plane nodes. See the Load balancing requirements for user-provisioned infrastructure section for more information.

  • When you create the Kubernetes manifest files in the following procedure, ensure that the mastersSchedulable parameter in the <installation_directory>/manifests/cluster-scheduler-02-config.yml file is set to true. This enables your application workloads to run on the control plane nodes.

  • Do not deploy any compute nodes when you create the Fedora CoreOS (FCOS) machines.

Creating the Kubernetes manifest and Ignition config files

Because you must modify some cluster definition files and manually start the cluster machines, you must generate the Kubernetes manifest and Ignition config files that the cluster needs to configure the machines.

The installation configuration file transforms into the Kubernetes manifests. The manifests wrap into the Ignition configuration files, which are later used to configure the cluster machines.

  • The Ignition config files that the OKD installation program generates contain certificates that expire after 24 hours, which are then renewed at that time. If the cluster is shut down before renewing the certificates and the cluster is later restarted after the 24 hours have elapsed, the cluster automatically recovers the expired certificates. The exception is that you must manually approve the pending node-bootstrapper certificate signing requests (CSRs) to recover kubelet certificates. See the documentation for Recovering from expired control plane certificates for more information.

  • It is recommended that you use Ignition config files within 12 hours after they are generated because the 24-hour certificate rotates from 16 to 22 hours after the cluster is installed. By using the Ignition config files within 12 hours, you can avoid installation failure if the certificate update runs during installation.

Prerequisites

  • You obtained the OKD installation program. For a restricted network installation, these files are on your mirror host.

  • You created the install-config.yaml installation configuration file.

Procedure

  1. Change to the directory that contains the OKD installation program and generate the Kubernetes manifests for the cluster:

    1. $ ./openshift-install create manifests --dir <installation_directory> (1)

    Example output

    1. INFO Credentials loaded from the "myprofile" profile in file "/home/myuser/.aws/credentials"
    2. INFO Consuming Install Config from target directory
    3. INFO Manifests created in: install_dir/manifests and install_dir/openshift
    1For <installation_directory>, specify the installation directory that contains the install-config.yaml file you created.

    If you are installing a three-node cluster, skip the following step to allow the control plane nodes to be schedulable.

    When you configure control plane nodes from the default unschedulable to schedulable, additional subscriptions are required. This is because control plane nodes then become worker nodes.

  2. Check that the mastersSchedulable parameter in the <installation_directory>/manifests/cluster-scheduler-02-config.yml Kubernetes manifest file is set to false. This setting prevents pods from being scheduled on the control plane machines:

    1. Open the <installation_directory>/manifests/cluster-scheduler-02-config.yml file.

    2. Locate the mastersSchedulable parameter and ensure that it is set to false.

    3. Save and exit the file.

  3. To create the Ignition configuration files, run the following command from the directory that contains the installation program:

    1. $ ./openshift-install create ignition-configs --dir <installation_directory> (1)
    1For <installation_directory>, specify the same installation directory.

    Ignition config files are created for the bootstrap, control plane, and compute nodes in the installation directory. The kubeadmin-password and kubeconfig files are created in the ./<installation_directory>/auth directory:

    1. .
    2. ├── auth
    3. ├── kubeadmin-password
    4. └── kubeconfig
    5. ├── bootstrap.ign
    6. ├── master.ign
    7. ├── metadata.json
    8. └── worker.ign

Additional resources

Configuring chrony time service

You must set the time server and related settings used by the chrony time service (chronyd) by modifying the contents of the chrony.conf file and passing those contents to your nodes as a machine config.

Procedure

  1. Create a Butane config including the contents of the chrony.conf file. For example, to configure chrony on worker nodes, create a 99-worker-chrony.bu file.

    See “Creating machine configs with Butane” for information about Butane.

    1. variant: openshift
    2. version: 4.10.0
    3. metadata:
    4. name: 99-worker-chrony (1)
    5. labels:
    6. machineconfiguration.openshift.io/role: worker (1)
    7. storage:
    8. files:
    9. - path: /etc/chrony.conf
    10. mode: 0644 (2)
    11. overwrite: true
    12. contents:
    13. inline: |
    14. pool 0.rhel.pool.ntp.org iburst (3)
    15. driftfile /var/lib/chrony/drift
    16. makestep 1.0 3
    17. rtcsync
    18. logdir /var/log/chrony
    1On control plane nodes, substitute master for worker in both of these locations.
    2Specify an octal value mode for the mode field in the machine config file. After creating the file and applying the changes, the mode is converted to a decimal value. You can check the YAML file with the command oc get mc <mc-name> -o yaml.
    3Specify any valid, reachable time source, such as the one provided by your DHCP server.
  2. Use Butane to generate a MachineConfig object file, 99-worker-chrony.yaml, containing the configuration to be delivered to the nodes:

    1. $ butane 99-worker-chrony.bu -o 99-worker-chrony.yaml
  3. Apply the configurations in one of two ways:

    • If the cluster is not running yet, after you generate manifest files, add the MachineConfig object file to the <installation_directory>/openshift directory, and then continue to create the cluster.

    • If the cluster is already running, apply the file:

      1. $ oc apply -f ./99-worker-chrony.yaml

Installing FCOS and starting the OKD bootstrap process

To install OKD on bare metal infrastructure that you provision, you must install Fedora CoreOS (FCOS) on the machines. When you install FCOS, you must provide the Ignition config file that was generated by the OKD installation program for the type of machine you are installing. If you have configured suitable networking, DNS, and load balancing infrastructure, the OKD bootstrap process begins automatically after the FCOS machines have rebooted.

To install FCOS on the machines, follow either the steps to use an ISO image or network PXE booting.

The compute node deployment steps included in this installation document are FCOS-specific. If you choose instead to deploy Fedora-based compute nodes, you take responsibility for all operating system life cycle management and maintenance, including performing system updates, applying patches, and completing all other required tasks. Only Fedora 8 compute machines are supported.

You can configure FCOS during ISO and PXE installations by using the following methods:

  • Kernel arguments: You can use kernel arguments to provide installation-specific information. For example, you can specify the locations of the FCOS installation files that you uploaded to your HTTP server and the location of the Ignition config file for the type of node you are installing. For a PXE installation, you can use the APPEND parameter to pass the arguments to the kernel of the live installer. For an ISO installation, you can interrupt the live installation boot process to add the kernel arguments. In both installation cases, you can use special coreos.inst.* arguments to direct the live installer, as well as standard installation boot arguments for turning standard kernel services on or off.

  • Ignition configs: OKD Ignition config files (*.ign) are specific to the type of node you are installing. You pass the location of a bootstrap, control plane, or compute node Ignition config file during the FCOS installation so that it takes effect on first boot. In special cases, you can create a separate, limited Ignition config to pass to the live system. That Ignition config could do a certain set of tasks, such as reporting success to a provisioning system after completing installation. This special Ignition config is consumed by the coreos-installer to be applied on first boot of the installed system. Do not provide the standard control plane and compute node Ignition configs to the live ISO directly.

  • coreos-installer: You can boot the live ISO installer to a shell prompt, which allows you to prepare the permanent system in a variety of ways before first boot. In particular, you can run the coreos-installer command to identify various artifacts to include, work with disk partitions, and set up networking. In some cases, you can configure features on the live system and copy them to the installed system.

Whether to use an ISO or PXE install depends on your situation. A PXE install requires an available DHCP service and more preparation, but can make the installation process more automated. An ISO install is a more manual process and can be inconvenient if you are setting up more than a few machines.

As of OKD 4.6, the FCOS ISO and other installation artifacts provide support for installation on disks with 4K sectors.

Installing FCOS by using an ISO image

You can use an ISO image to install FCOS on the machines.

Prerequisites

  • You have created the Ignition config files for your cluster.

  • You have configured suitable network, DNS and load balancing infrastructure.

  • You have an HTTP server that can be accessed from your computer, and from the machines that you create.

  • You have reviewed the Advanced FCOS installation configuration section for different ways to configure features, such as networking and disk partitioning.

Procedure

  1. Obtain the SHA512 digest for each of your Ignition config files. For example, you can use the following on a system running Linux to get the SHA512 digest for your bootstrap.ign Ignition config file:

    1. $ sha512sum <installation_directory>/bootstrap.ign

    Example output

    1. a5a2d43879223273c9b60af66b44202a1d1248fc01cf156c46d4a79f552b6bad47bc8cc78ddf0116e80c59d2ea9e32ba53bc807afbca581aa059311def2c3e3b installation_directory/bootstrap.ign

    The digests are provided to the coreos-installer in a later step to validate the authenticity of the Ignition config files on the cluster nodes.

  2. Upload the bootstrap, control plane, and compute node Ignition config files that the installation program created to your HTTP server. Note the URLs of these files.

    You can add or change configuration settings in your Ignition configs before saving them to your HTTP server. If you plan to add more compute machines to your cluster after you finish installation, do not delete these files.

  3. From the installation host, validate that the Ignition config files are available on the URLs. The following example gets the Ignition config file for the bootstrap node:

    1. $ curl -k http://<HTTP_server>/bootstrap.ign (1)

    Example output

    1. % Total % Received % Xferd Average Speed Time Time Time Current
    2. Dload Upload Total Spent Left Speed
    3. 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0{"ignition":{"version":"3.2.0"},"passwd":{"users":[{"name":"core","sshAuthorizedKeys":["ssh-rsa...

    Replace bootstrap.ign with master.ign or worker.ign in the command to validate that the Ignition config files for the control plane and compute nodes are also available.

  4. Obtain the FCOS images from the FCOS Downloads page

  5. Use the ISO to start the FCOS installation. Use one of the following installation options:

    • Burn the ISO image to a disk and boot it directly.

    • Use ISO redirection by using a lights-out management (LOM) interface.

  6. Boot the FCOS ISO image without specifying any options or interrupting the live boot sequence. Wait for the installer to boot into a shell prompt in the FCOS live environment.

    It is possible to interrupt the FCOS installation boot process to add kernel arguments. However, for this ISO procedure you should use the coreos-installer command as outlined in the following steps, instead of adding kernel arguments.

  7. Run the coreos-installer command and specify the options that meet your installation requirements. At a minimum, you must specify the URL that points to the Ignition config file for the node type, and the device that you are installing to:

    1. $ sudo coreos-installer install --ignition-url=http://<HTTP_server>/<node_type>.ign <device> --ignition-hash=SHA512-<digest> (1) (2)
    1You must run the coreos-installer command by using sudo, because the core user does not have the required root privileges to perform the installation.
    2The —ignition-hash option is required when the Ignition config file is obtained through an HTTP URL to validate the authenticity of the Ignition config file on the cluster node. <digest> is the Ignition config file SHA512 digest obtained in a preceding step.

    If you want to provide your Ignition config files through an HTTPS server that uses TLS, you can add the internal certificate authority (CA) to the system trust store before running coreos-installer.

    The following example initializes a bootstrap node installation to the /dev/sda device. The Ignition config file for the bootstrap node is obtained from an HTTP web server with the IP address 192.168.1.2:

    1. $ sudo coreos-installer install --ignition-url=http://192.168.1.2:80/installation_directory/bootstrap.ign /dev/sda --ignition-hash=SHA512-a5a2d43879223273c9b60af66b44202a1d1248fc01cf156c46d4a79f552b6bad47bc8cc78ddf0116e80c59d2ea9e32ba53bc807afbca581aa059311def2c3e3b
  8. Monitor the progress of the FCOS installation on the console of the machine.

    Be sure that the installation is successful on each node before commencing with the OKD installation. Observing the installation process can also help to determine the cause of FCOS installation issues that might arise.

  9. After FCOS installs, the system reboots. During the system reboot, it applies the Ignition config file that you specified.

  10. Check the console output to verify that Ignition ran.

    Example command

    1. Ignition: ran on 2022/03/14 14:48:33 UTC (this boot)
    2. Ignition: user-provided config was applied
  11. Continue to create the other machines for your cluster.

    You must create the bootstrap and control plane machines at this time. If the control plane machines are not made schedulable, also create at least two compute machines before you install OKD.

    If the required network, DNS, and load balancer infrastructure are in place, the OKD bootstrap process begins automatically after the FCOS nodes have rebooted.

    FCOS nodes do not include a default password for the core user. You can access the nodes by running ssh core@<node>.<cluster_name>.<base_domain> as a user with access to the SSH private key that is paired to the public key that you specified in your install_config.yaml file. OKD 4 cluster nodes running FCOS are immutable and rely on Operators to apply cluster changes. Accessing cluster nodes by using SSH is not recommended. However, when investigating installation issues, if the OKD API is not available, or the kubelet is not properly functioning on a target node, SSH access might be required for debugging or disaster recovery.

Installing FCOS by using PXE or iPXE booting

You can use PXE or iPXE booting to install FCOS on the machines.

Prerequisites

  • You have created the Ignition config files for your cluster.

  • You have configured suitable network, DNS and load balancing infrastructure.

  • You have configured suitable PXE or iPXE infrastructure.

  • You have an HTTP server that can be accessed from your computer, and from the machines that you create.

  • You have reviewed the Advanced FCOS installation configuration section for different ways to configure features, such as networking and disk partitioning.

Procedure

  1. Upload the bootstrap, control plane, and compute node Ignition config files that the installation program created to your HTTP server. Note the URLs of these files.

    You can add or change configuration settings in your Ignition configs before saving them to your HTTP server. If you plan to add more compute machines to your cluster after you finish installation, do not delete these files.

  2. From the installation host, validate that the Ignition config files are available on the URLs. The following example gets the Ignition config file for the bootstrap node:

    1. $ curl -k http://<HTTP_server>/bootstrap.ign (1)

    Example output

    1. % Total % Received % Xferd Average Speed Time Time Time Current
    2. Dload Upload Total Spent Left Speed
    3. 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0{"ignition":{"version":"3.2.0"},"passwd":{"users":[{"name":"core","sshAuthorizedKeys":["ssh-rsa...

    Replace bootstrap.ign with master.ign or worker.ign in the command to validate that the Ignition config files for the control plane and compute nodes are also available.

  3. Obtain the FCOS kernel, initramfs and rootfs files from the FCOS Downloads page

  4. Upload the rootfs, kernel, and initramfs files to your HTTP server.

    If you plan to add more compute machines to your cluster after you finish installation, do not delete these files.

  5. Configure the network boot infrastructure so that the machines boot from their local disks after FCOS is installed on them.

  6. Configure PXE or iPXE installation for the FCOS images and begin the installation.

    Modify one of the following example menu entries for your environment and verify that the image and Ignition files are properly accessible:

    • For PXE (x86_64):

      1. DEFAULT pxeboot
      2. TIMEOUT 20
      3. PROMPT 0
      4. LABEL pxeboot
      5. KERNEL http://<HTTP_server>/rhcos-<version>-live-kernel-<architecture> (1)
      6. APPEND initrd=http://<HTTP_server>/rhcos-<version>-live-initramfs.<architecture>.img coreos.live.rootfs_url=http://<HTTP_server>/rhcos-<version>-live-rootfs.<architecture>.img coreos.inst.install_dev=/dev/sda coreos.inst.ignition_url=http://<HTTP_server>/bootstrap.ign (2) (3)
      1Specify the location of the live kernel file that you uploaded to your HTTP server. The URL must be HTTP, TFTP, or FTP; HTTPS and NFS are not supported.
      2If you use multiple NICs, specify a single interface in the ip option. For example, to use DHCP on a NIC that is named eno1, set ip=eno1:dhcp.
      3Specify the locations of the FCOS files that you uploaded to your HTTP server. The initrd parameter value is the location of the initramfs file, the coreos.live.rootfs_url parameter value is the location of the rootfs file, and the coreos.inst.ignition_url parameter value is the location of the bootstrap Ignition config file. You can also add more kernel arguments to the APPEND line to configure networking or other boot options.

      This configuration does not enable serial console access on machines with a graphical console. To configure a different console, add one or more console= arguments to the APPEND line. For example, add console=tty0 console=ttyS0 to set the first PC serial port as the primary console and the graphical console as a secondary console. For more information, see How does one set up a serial terminal and/or console in Red Hat Enterprise Linux?.

    • For iPXE (x86_64 + arm64):

      1. kernel http://<HTTP_server>/rhcos-<version>-live-kernel-<architecture> initrd=main coreos.live.rootfs_url=http://<HTTP_server>/rhcos-<version>-live-rootfs.<architecture>.img coreos.inst.install_dev=/dev/sda coreos.inst.ignition_url=http://<HTTP_server>/bootstrap.ign (1) (2)
      2. initrd --name main http://<HTTP_server>/rhcos-<version>-live-initramfs.<architecture>.img (3)
      3. boot
      1Specify the locations of the FCOS files that you uploaded to your HTTP server. The kernel parameter value is the location of the kernel file, the initrd=main argument is needed for booting on UEFI systems, the coreos.live.rootfs_url parameter value is the location of the rootfs file, and the coreos.inst.ignition_url parameter value is the location of the bootstrap Ignition config file.
      2If you use multiple NICs, specify a single interface in the ip option. For example, to use DHCP on a NIC that is named eno1, set ip=eno1:dhcp.
      3Specify the location of the initramfs file that you uploaded to your HTTP server.

      This configuration does not enable serial console access on machines with a graphical console. To configure a different console, add one or more console= arguments to the kernel line. For example, add console=tty0 console=ttyS0 to set the first PC serial port as the primary console and the graphical console as a secondary console. For more information, see How does one set up a serial terminal and/or console in Red Hat Enterprise Linux?.

      To network boot the CoreOS kernel on arm64 architecture, you need to use a version of iPXE build with the IMAGE_GZIP option enabled. See IMAGE_GZIP option in iPXE.

    • For PXE (with UEFI and Grub as second stage) on arm64:

      1. menuentry 'Install CoreOS' {
      2. linux rhcos-<version>-live-kernel-<architecture> coreos.live.rootfs_url=http://<HTTP_server>/rhcos-<version>-live-rootfs.<architecture>.img coreos.inst.install_dev=/dev/sda coreos.inst.ignition_url=http://<HTTP_server>/bootstrap.ign (1) (2)
      3. initrd rhcos-<version>-live-initramfs.<architecture>.img (3)
      4. }
      1Specify the locations of the FCOS files that you uploaded to your HTTP/TFTP server. The kernel parameter value is the location of the kernel file on your TFTP server. The coreos.live.rootfs_url parameter value is the location of the rootfs file, and the coreos.inst.ignition_url parameter value is the location of the bootstrap Ignition config file on your HTTP Server.
      2If you use multiple NICs, specify a single interface in the ip option. For example, to use DHCP on a NIC that is named eno1, set ip=eno1:dhcp.
      3Specify the location of the initramfs file that you uploaded to your TFTP server.
  7. Monitor the progress of the FCOS installation on the console of the machine.

    Be sure that the installation is successful on each node before commencing with the OKD installation. Observing the installation process can also help to determine the cause of FCOS installation issues that might arise.

  8. After FCOS installs, the system reboots. During reboot, the system applies the Ignition config file that you specified.

  9. Check the console output to verify that Ignition ran.

    Example command

    1. Ignition: ran on 2022/03/14 14:48:33 UTC (this boot)
    2. Ignition: user-provided config was applied
  10. Continue to create the machines for your cluster.

    You must create the bootstrap and control plane machines at this time. If the control plane machines are not made schedulable, also create at least two compute machines before you install the cluster.

    If the required network, DNS, and load balancer infrastructure are in place, the OKD bootstrap process begins automatically after the FCOS nodes have rebooted.

    FCOS nodes do not include a default password for the core user. You can access the nodes by running ssh core@<node>.<cluster_name>.<base_domain> as a user with access to the SSH private key that is paired to the public key that you specified in your install_config.yaml file. OKD 4 cluster nodes running FCOS are immutable and rely on Operators to apply cluster changes. Accessing cluster nodes by using SSH is not recommended. However, when investigating installation issues, if the OKD API is not available, or the kubelet is not properly functioning on a target node, SSH access might be required for debugging or disaster recovery.

Advanced FCOS installation configuration

A key benefit for manually provisioning the Fedora CoreOS (FCOS) nodes for OKD is to be able to do configuration that is not available through default OKD installation methods. This section describes some of the configurations that you can do using techniques that include:

  • Passing kernel arguments to the live installer

  • Running coreos-installer manually from the live system

  • Customizing a live ISO or PXE boot image

The advanced configuration topics for manual Fedora CoreOS (FCOS) installations detailed in this section relate to disk partitioning, networking, and using Ignition configs in different ways.

Using advanced networking options for PXE and ISO installations

Networking for OKD nodes uses DHCP by default to gather all necessary configuration settings. To set up static IP addresses or configure special settings, such as bonding, you can do one of the following:

  • Pass special kernel parameters when you boot the live installer.

  • Use a machine config to copy networking files to the installed system.

  • Configure networking from a live installer shell prompt, then copy those settings to the installed system so that they take effect when the installed system first boots.

To configure a PXE or iPXE installation, use one of the following options:

  • See the “Advanced RHCOS installation reference” tables.

  • Use a machine config to copy networking files to the installed system.

To configure an ISO installation, use the following procedure.

Procedure

  1. Boot the ISO installer.

  2. From the live system shell prompt, configure networking for the live system using available RHEL tools, such as nmcli or nmtui.

  3. Run the coreos-installer command to install the system, adding the --copy-network option to copy networking configuration. For example:

    1. $ coreos-installer install --copy-network \
    2. --ignition-url=http://host/worker.ign /dev/sda

    The —copy-network option only copies networking configuration found under /etc/NetworkManager/system-connections. In particular, it does not copy the system hostname.

  4. Reboot into the installed system.

Additional resources

Disk partitioning

The disk partitions are created on OKD cluster nodes during the Fedora CoreOS (FCOS) installation. Each FCOS node of a particular architecture uses the same partition layout, unless the default partitioning configuration is overridden. During the FCOS installation, the size of the root file system is increased to use the remaining available space on the target device.

There are two cases where you might want to override the default partitioning when installing FCOS on an OKD cluster node:

  • Creating separate partitions: For greenfield installations on an empty disk, you might want to add separate storage to a partition. This is officially supported for mounting /var or a subdirectory of /var, such as /var/lib/etcd, on a separate partition, but not both.

    Kubernetes supports only two file system partitions. If you add more than one partition to the original configuration, Kubernetes cannot monitor all of them.

  • Retaining existing partitions: For a brownfield installation where you are reinstalling OKD on an existing node and want to retain data partitions installed from your previous operating system, there are both boot arguments and options to coreos-installer that allow you to retain existing data partitions.

The use of custom partitions could result in those partitions not being monitored by OKD or alerted on. If you are overriding the default partitioning, see Understanding OpenShift File System Monitoring (eviction conditions) for more information about how OKD monitors your host file systems.

Creating a separate /var partition

In general, you should use the default disk partitioning that is created during the FCOS installation. However, there are cases where you might want to create a separate partition for a directory that you expect to grow.

OKD supports the addition of a single partition to attach storage to either the /var directory or a subdirectory of /var. For example:

  • /var/lib/containers: Holds container-related content that can grow as more images and containers are added to a system.

  • /var/lib/etcd: Holds data that you might want to keep separate for purposes such as performance optimization of etcd storage.

  • /var: Holds data that you might want to keep separate for purposes such as auditing.

Storing the contents of a /var directory separately makes it easier to grow storage for those areas as needed and reinstall OKD at a later date and keep that data intact. With this method, you will not have to pull all your containers again, nor will you have to copy massive log files when you update systems.

The use of a separate partition for the /var directory or a subdirectory of /var also prevents data growth in the partitioned directory from filling up the root file system.

The following procedure sets up a separate /var partition by adding a machine config manifest that is wrapped into the Ignition config file for a node type during the preparation phase of an installation.

Procedure

  1. On your installation host, change to the directory that contains the OKD installation program and generate the Kubernetes manifests for the cluster:

    1. $ openshift-install create manifests --dir <installation_directory>
  2. Create a Butane config that configures the additional partition. For example, name the file $HOME/clusterconfig/98-var-partition.bu, change the disk device name to the name of the storage device on the worker systems, and set the storage size as appropriate. This example places the /var directory on a separate partition:

    1. variant: openshift
    2. version: 4.10.0
    3. metadata:
    4. labels:
    5. machineconfiguration.openshift.io/role: worker
    6. name: 98-var-partition
    7. storage:
    8. disks:
    9. - device: /dev/<device_name> (1)
    10. partitions:
    11. - label: var
    12. start_mib: <partition_start_offset> (2)
    13. size_mib: <partition_size> (3)
    14. filesystems:
    15. - device: /dev/disk/by-partlabel/var
    16. path: /var
    17. format: xfs
    18. mount_options: [defaults, prjquota] (4)
    19. with_mount_unit: true
    1The storage device name of the disk that you want to partition.
    2When adding a data partition to the boot disk, a minimum offset value of 25000 mebibytes is recommended. The root file system is automatically resized to fill all available space up to the specified offset. If no offset value is specified, or if the specified value is smaller than the recommended minimum, the resulting root file system will be too small, and future reinstalls of FCOS might overwrite the beginning of the data partition.
    3The size of the data partition in mebibytes.
    4The prjquota mount option must be enabled for filesystems used for container storage.

    When creating a separate /var partition, you cannot use different instance types for compute nodes, if the different instance types do not have the same device name.

  3. Create a manifest from the Butane config and save it to the clusterconfig/openshift directory. For example, run the following command:

    1. $ butane $HOME/clusterconfig/98-var-partition.bu -o $HOME/clusterconfig/openshift/98-var-partition.yaml
  4. Create the Ignition config files:

    1. $ openshift-install create ignition-configs --dir <installation_directory> (1)
    1For <installation_directory>, specify the same installation directory.

    Ignition config files are created for the bootstrap, control plane, and compute nodes in the installation directory:

    1. .
    2. ├── auth
    3. ├── kubeadmin-password
    4. └── kubeconfig
    5. ├── bootstrap.ign
    6. ├── master.ign
    7. ├── metadata.json
    8. └── worker.ign

    The files in the <installation_directory>/manifest and <installation_directory>/openshift directories are wrapped into the Ignition config files, including the file that contains the 98-var-partition custom MachineConfig object.

Next steps

  • You can apply the custom disk partitioning by referencing the Ignition config files during the FCOS installations.
Retaining existing partitions

For an ISO installation, you can add options to the coreos-installer command that cause the installer to maintain one or more existing partitions. For a PXE installation, you can add coreos.inst.* options to the APPEND parameter to preserve partitions.

Saved partitions might be data partitions from an existing OKD system. You can identify the disk partitions you want to keep either by partition label or by number.

If you save existing partitions, and those partitions do not leave enough space for FCOS, the installation will fail without damaging the saved partitions.

Retaining existing partitions during an ISO installation

This example preserves any partition in which the partition label begins with data (data*):

  1. # coreos-installer install --ignition-url http://10.0.2.2:8080/user.ign \
  2. --save-partlabel 'data*' /dev/sda

The following example illustrates running the coreos-installer in a way that preserves the sixth (6) partition on the disk:

  1. # coreos-installer install --ignition-url http://10.0.2.2:8080/user.ign \
  2. --save-partindex 6 /dev/sda

This example preserves partitions 5 and higher:

  1. # coreos-installer install --ignition-url http://10.0.2.2:8080/user.ign
  2. --save-partindex 5- /dev/sda

In the previous examples where partition saving is used, coreos-installer recreates the partition immediately.

Retaining existing partitions during a PXE installation

This APPEND option preserves any partition in which the partition label begins with ‘data’ (‘data*‘):

  1. coreos.inst.save_partlabel=data*

This APPEND option preserves partitions 5 and higher:

  1. coreos.inst.save_partindex=5-

This APPEND option preserves partition 6:

  1. coreos.inst.save_partindex=6

Identifying Ignition configs

When doing an FCOS manual installation, there are two types of Ignition configs that you can provide, with different reasons for providing each one:

  • Permanent install Ignition config: Every manual FCOS installation needs to pass one of the Ignition config files generated by openshift-installer, such as bootstrap.ign, master.ign and worker.ign, to carry out the installation.

    It is not recommended to modify these Ignition config files directly. You can update the manifest files that are wrapped into the Ignition config files, as outlined in examples in the preceding sections.

    For PXE installations, you pass the Ignition configs on the APPEND line using the coreos.inst.ignition_url= option. For ISO installations, after the ISO boots to the shell prompt, you identify the Ignition config on the coreos-installer command line with the --ignition-url= option. In both cases, only HTTP and HTTPS protocols are supported.

  • Live install Ignition config: This type can be created by using the coreos-installer customize subcommand and its various options. With this method, the Ignition config passes to the live install medium, runs immediately upon booting, and performs setup tasks before or after the FCOS system installs to disk. This method should only be used for performing tasks that must be done once and not applied again later, such as with advanced partitioning that cannot be done using a machine config.

    For PXE or ISO boots, you can create the Ignition config and APPEND the ignition.config.url= option to identify the location of the Ignition config. You also need to append ignition.firstboot ignition.platform.id=metal or the ignition.config.url option will be ignored.

Customizing a live FCOS ISO or PXE install

You can use the live ISO image or PXE environment to install FCOS by injecting an Ignition config file directly into the image. This creates a customized image that you can use to provision your system.

For an ISO image, the mechanism to do this is the coreos-installer iso customize subcommand, which modifies the .iso file with your configuration. Similarly, the mechanism for a PXE environment is the coreos-installer pxe customize subcommand, which creates a new initramfs file that includes your customizations.

The customize subcommand is a general purpose tool that can embed other types of customizations as well. The following tasks are examples of some of the more common customizations:

  • Inject custom CA certificates for when corporate security policy requires their use.

  • Configure network settings without the need for kernel arguments.

  • Embed arbitrary pre-install and post-install scripts or binaries.

Customizing a live FCOS ISO image

You can customize a live FCOS ISO image directly with the coreos-installer iso customize subcommand. When you boot the ISO image, the customizations are applied automatically.

You can use this feature to configure the ISO image to automatically install FCOS.

Procedure

  1. Download the coreos-installer binary from the coreos-installer image mirror page.

  2. Retrieve the FCOS ISO image from the FCOS image mirror page and the Ignition config file, and then run the following command to inject the Ignition config directly into the ISO image:

    1. $ coreos-installer iso customize rhcos-<version>-live.x86_64.iso \
    2. --dest-ignition bootstrap.ign \ (1)
    3. --dest-device /dev/sda (2)
1The Ignition config file that is generated from openshift-installer.
2When you specify this option, the ISO image automatically runs an install. Otherwise, the image remains configured for installing, but does not do so automatically unless you specify the coreos.inst.install_dev kernel argument.

Your customizations are applied and affect every subsequent boot of the ISO image.

  1. To remove the ISO image customizations and return the image to its pristine state, run:

    1. $ coreos-installer iso reset rhcos-<version>-live.x86_64.iso

    You can now re-customize the live ISO image or use it in its pristine state.

Modifying a live install ISO image to use a custom certificate authority

You can provide certificate authority (CA) certificates to Ignition with the --ignition-ca flag of the customize subcommand. You can use the CA certificates during both the installation boot and when provisioning the installed system.

Procedure

  1. Download the coreos-installer binary from the coreos-installer image mirror page.

  2. Retrieve the FCOS ISO image from the FCOS image mirror page and run the following command to customize the ISO image for use with a custom CA:

    1. $ coreos-installer iso customize rhcos-<version>-live.x86_64.iso --ignition-ca cert.pem

    Custom CA certificates affect how Ignition fetches remote resources but they do not affect the certificates installed onto the system.

    Your CA certificate is applied and affects every subsequent boot of the ISO image.

Modifying a live install ISO image with customized network settings

You can embed a NetworkManager keyfile into the live ISO image and pass it through to the installed system with the --network-keyfile flag of the customize subcommand.

Procedure

  1. Download the coreos-installer binary from the coreos-installer image mirror page.

  2. Create a connection profile for a bonded interface. For example, create the bond0.nmconnection file in your local directory with the following content:

    1. [connection]
    2. id=bond0
    3. type=bond
    4. interface-name=bond0
    5. multi-connect=1
    6. permissions=
    7. [ethernet]
    8. mac-address-blacklist=
    9. [bond]
    10. miimon=100
    11. mode=active-backup
    12. [ipv4]
    13. method=auto
    14. [ipv6]
    15. method=auto
    16. [proxy]
  3. Create a connection profile for a secondary interface to add to the bond. For example, create the bond0-proxy-em1.nmconnection file in your local directory with the following content:

    1. [connection]
    2. id=em1
    3. type=ethernet
    4. interface-name=em1
    5. master=bond0
    6. multi-connect=1
    7. permissions=
    8. slave-type=bond
    9. [ethernet]
    10. mac-address-blacklist=
  4. Create a connection profile for a secondary interface to add to the bond. For example, create the bond0-proxy-em2.nmconnection file in your local directory with the following content:

    1. [connection]
    2. id=em2
    3. type=ethernet
    4. interface-name=em2
    5. master=bond0
    6. multi-connect=1
    7. permissions=
    8. slave-type=bond
    9. [ethernet]
    10. mac-address-blacklist=
  5. Retrieve the FCOS ISO image from the FCOS image mirror page and run the following command to customize the ISO image with your configured networking:

    1. $ coreos-installer iso customize rhcos-<version>-live.x86_64.iso \
    2. --network-keyfile bond0.nmconnection \
    3. --network-keyfile bond0-proxy-em1.nmconnection \
    4. --network-keyfile bond0-proxy-em2.nmconnection

    Network settings are applied to the live system and are carried over to the destination system.

Customizing a live FCOS PXE environment

You can customize a live FCOS PXE environment directly with the coreos-installer pxe customize subcommand. When you boot the PXE environment, the customizations are applied automatically.

You can use this feature to configure the PXE environment to automatically install FCOS.

Procedure

  1. Download the coreos-installer binary from the coreos-installer image mirror page.

  2. Retrieve the FCOS kernel, initramfs and rootfs files from the FCOS image mirror page and the Ignition config file, and then run the following command to create a new initramfs file that contains the customizations from your Ignition config:

    1. $ coreos-installer pxe customize rhcos-<version>-live-initramfs.x86_64.img \
    2. --dest-ignition bootstrap.ign \ (1)
    3. --dest-device /dev/sda \ (2)
    4. -o rhcos-<version>-custom-initramfs.x86_64.img
    1The Ignition config file that is generated from openshift-installer.
    2When you specify this option, the PXE environment automatically runs an install. Otherwise, the image remains configured for installing, but does not do so automatically unless you specify the coreos.inst.install_dev kernel argument.

    Your customizations are applied and affect every subsequent boot of the PXE environment.

Modifying a live install PXE environment to use a custom certificate authority

You can provide certificate authority (CA) certificates to Ignition with the --ignition-ca flag of the customize subcommand. You can use the CA certificates during both the installation boot and when provisioning the installed system.

Procedure

  1. Download the coreos-installer binary from the coreos-installer image mirror page.

  2. Retrieve the FCOS kernel, initramfs and rootfs files from the FCOS image mirror page and run the following command to create a new customized initramfs file for use with a custom CA:

    1. $ coreos-installer pxe customize rhcos-<version>-live-initramfs.x86_64.img \
    2. --ignition-ca cert.pem \
    3. -o rhcos-<version>-custom-initramfs.x86_64.img

    Custom CA certificates affect how Ignition fetches remote resources but they do not affect the certificates installed onto the system.

    Your CA certificate is applied and affects every subsequent boot of the PXE environment.

Modifying a live install PXE environment with customized network settings

You can embed a NetworkManager keyfile into the live PXE environment and pass it through to the installed system with the --network-keyfile flag of the customize subcommand.

Procedure

  1. Download the coreos-installer binary from the coreos-installer image mirror page.

  2. Create a connection profile for a bonded interface. For example, create the bond0.nmconnection file in your local directory with the following content:

    1. [connection]
    2. id=bond0
    3. type=bond
    4. interface-name=bond0
    5. multi-connect=1
    6. permissions=
    7. [ethernet]
    8. mac-address-blacklist=
    9. [bond]
    10. miimon=100
    11. mode=active-backup
    12. [ipv4]
    13. method=auto
    14. [ipv6]
    15. method=auto
    16. [proxy]
  3. Create a connection profile for a secondary interface to add to the bond. For example, create the bond0-proxy-em1.nmconnection file in your local directory with the following content:

    1. [connection]
    2. id=em1
    3. type=ethernet
    4. interface-name=em1
    5. master=bond0
    6. multi-connect=1
    7. permissions=
    8. slave-type=bond
    9. [ethernet]
    10. mac-address-blacklist=
  4. Create a connection profile for a secondary interface to add to the bond. For example, create the bond0-proxy-em2.nmconnection file in your local directory with the following content:

    1. [connection]
    2. id=em2
    3. type=ethernet
    4. interface-name=em2
    5. master=bond0
    6. multi-connect=1
    7. permissions=
    8. slave-type=bond
    9. [ethernet]
    10. mac-address-blacklist=
  5. Retrieve the FCOS kernel, initramfs and rootfs files from the FCOS image mirror page and run the following command to create a new customized initramfs file that contains your configured networking:

    1. $ coreos-installer pxe customize rhcos-<version>-live-initramfs.x86_64.img \
    2. --network-keyfile bond0.nmconnection \
    3. --network-keyfile bond0-proxy-em1.nmconnection \
    4. --network-keyfile bond0-proxy-em2.nmconnection \
    5. -o rhcos-<version>-custom-initramfs.x86_64.img

    Network settings are applied to the live system and are carried over to the destination system.

Advanced FCOS installation reference

This section illustrates the networking configuration and other advanced options that allow you to modify the Fedora CoreOS (FCOS) manual installation process. The following tables describe the kernel arguments and command-line options you can use with the FCOS live installer and the coreos-installer command.

Networking and bonding options for ISO installations

If you install FCOS from an ISO image, you can add kernel arguments manually when you boot the image to configure networking for a node. If no networking arguments are specified, DHCP is activated in the initramfs when FCOS detects that networking is required to fetch the Ignition config file.

When adding networking arguments manually, you must also add the rd.neednet=1 kernel argument to bring the network up in the initramfs.

The following table provides examples for configuring networking and bonding on your FCOS nodes for ISO installations. The examples describe how to use the ip=, nameserver=, and bond= kernel arguments.

Ordering is important when adding the kernel arguments: ip=, nameserver=, and then bond=.

The networking options are passed to the dracut tool during system boot. For more information about the networking options supported by dracut, see the dracut.cmdline manual page.

Table 11. Networking and bonding options for ISO installations
DescriptionExamples

To configure an IP address, either use DHCP (ip=dhcp) or set an individual static IP address (ip=<host_ip>). If setting a static IP, you must then identify the DNS server IP address (nameserver=<dns_ip>) on each node. This example sets:

  • The node’s IP address to 10.10.10.2

  • The gateway address to 10.10.10.254

  • The netmask to 255.255.255.0

  • The hostname to core0.example.com

  • The DNS server address to 4.4.4.41

  • The auto-configuration value to none. No auto-configuration is required when IP networking is configured statically.

When you use DHCP to configure IP addressing for the FCOS machines, the machines also obtain the DNS server information through DHCP. For DHCP-based deployments, you can define the DNS server address that is used by the FCOS nodes through your DHCP server configuration.

  1. ip=10.10.10.2::10.10.10.254:255.255.255.0:core0.example.com:enp1s0:none
  2. nameserver=4.4.4.41

Specify multiple network interfaces by specifying multiple ip= entries.

  1. ip=10.10.10.2::10.10.10.254:255.255.255.0:core0.example.com:enp1s0:none
  2. ip=10.10.10.3::10.10.10.254:255.255.255.0:core0.example.com:enp2s0:none

Optional: You can configure routes to additional networks by setting an rd.route= value.

When you configure one or multiple networks, one default gateway is required. If the additional network gateway is different from the primary network gateway, the default gateway must be the primary network gateway.

To configure the default gateway:

  1. ip=::10.10.10.254::::

To configure the route for the additional network:

  1. rd.route=20.20.20.0/24:20.20.20.254:enp2s0

Disable DHCP on a single interface, such as when there are two or more network interfaces and only one interface is being used. In the example, the enp1s0 interface has a static networking configuration and DHCP is disabled for enp2s0, which is not used.

  1. ip=10.10.10.2::10.10.10.254:255.255.255.0:core0.example.com:enp1s0:none
  2. ip=::::core0.example.com:enp2s0:none

You can combine DHCP and static IP configurations on systems with multiple network interfaces.

  1. ip=enp1s0:dhcp
  2. ip=10.10.10.2::10.10.10.254:255.255.255.0:core0.example.com:enp2s0:none

Optional: You can configure VLANs on individual interfaces by using the vlan= parameter.

To configure a VLAN on a network interface and use a static IP address:

  1. ip=10.10.10.2::10.10.10.254:255.255.255.0:core0.example.com:enp2s0.100:none
  2. vlan=enp2s0.100:enp2s0

To configure a VLAN on a network interface and to use DHCP:

  1. ip=enp2s0.100:dhcp
  2. vlan=enp2s0.100:enp2s0

You can provide multiple DNS servers by adding a nameserver= entry for each server.

  1. nameserver=1.1.1.1
  2. nameserver=8.8.8.8

Optional: Bonding multiple network interfaces to a single interface is supported using the bond= option. In these two examples:

  • The syntax for configuring a bonded interface is: bond=name[:network_interfaces][:options]

    name is the bonding device name (bond0), network_interfaces represents a comma-separated list of physical (ethernet) interfaces (em1,em2), and options is a comma-separated list of bonding options. Enter modinfo bonding to see available options.

  • When you create a bonded interface using bond=, you must specify how the IP address is assigned and other information for the bonded interface.

To configure the bonded interface to use DHCP, set the bond’s IP address to dhcp. For example:

  1. bond=bond0:em1,em2:mode=active-backup
  2. ip=bond0:dhcp

To configure the bonded interface to use a static IP address, enter the specific IP address you want and related information. For example:

  1. bond=bond0:em1,em2:mode=active-backup
  2. ip=10.10.10.2::10.10.10.254:255.255.255.0:core0.example.com:bond0:none

Optional: You can configure VLANs on bonded interfaces by using the vlan= parameter.

To configure the bonded interface with a VLAN and to use DHCP:

  1. ip=bond0.100:dhcp
  2. bond=bond0:em1,em2:mode=active-backup
  3. vlan=bond0.100:bond0

To configure the bonded interface with a VLAN and to use a static IP address:

  1. ip=10.10.10.2::10.10.10.254:255.255.255.0:core0.example.com:bond0.100:none
  2. bond=bond0:em1,em2:mode=active-backup
  3. vlan=bond0.100:bond0

Optional: Network teaming can be used as an alternative to bonding by using the team= parameter. In this example:

  • The syntax for configuring a team interface is: team=name[:network_interfaces]

    name is the team device name (team0) and network_interfaces represents a comma-separated list of physical (ethernet) interfaces (em1, em2).

Teaming is planned to be deprecated when FCOS switches to an upcoming version of Fedora. For more information, see this Red Hat Knowledgebase Article.

To configure a network team:

  1. team=team0:em1,em2
  2. ip=team0:dhcp
coreos-installer options for ISO and PXE installations

You can install FCOS by running coreos-installer install <options> <device> at the command prompt, after booting into the FCOS live environment from an ISO image.

The following table shows the subcommands, options, and arguments you can pass to the coreos-installer command.

Table 12. coreos-installer subcommands, command-line options, and arguments

coreos-installer install subcommand

Subcommand

Description

$ coreos-installer install <options> <device>

Embed an Ignition config in an ISO image.

coreos-installer install subcommand options

Option

Description

-u, —image-url <url>

Specify the image URL manually.

-f, —image-file <path>

Specify a local image file manually. Used for debugging.

-i, —ignition-file <path>

Embed an Ignition config from a file.

-I, —ignition-url <URL>

Embed an Ignition config from a URL.

—ignition-hash <digest>

Digest type-value of the Ignition config.

-p, —platform <name>

Override the Ignition platform ID for the installed system.

—append-karg <arg>…​

Append a default kernel argument to the installed system.

—delete-karg <arg>…​

Delete a default kernel argument from the installed system.

-n, —copy-network

Copy the network configuration from the install environment.

The —copy-network option only copies networking configuration found under /etc/NetworkManager/system-connections. In particular, it does not copy the system hostname.

—network-dir <path>

For use with -n. Default is /etc/NetworkManager/system-connections/.

—save-partlabel <lx>..

Save partitions with this label glob.

—save-partindex <id>…​

Save partitions with this number or range.

—insecure

Skip signature verification.

—insecure-ignition

Allow Ignition URL without HTTPS or hash.

—architecture <name>

Target CPU architecture. Valid values are x86_64 and arm64.

—preserve-on-error

Do not clear partition table on error.

-h, —help

Print help information.

coreos-installer install subcommand argument

Argument

Description

<device>

The destination device.

coreos-installer ISO subcommands

Subcommand

Description

$ coreos-installer iso customize <options> <ISO_image>

Customize a FCOS live ISO image.

coreos-installer iso reset <options> <ISO_image>

Restore a FCOS live ISO image to default settings.

coreos-installer iso ignition remove <options> <ISO_image>

Remove the embedded Ignition config from an ISO image.

coreos-installer ISO customize subcommand options

Option

Description

—dest-ignition <path>

Merge the specified Ignition config file into a new configuration fragment for the destination system.

—dest-device <path>

Install and overwrite the specified destination device.

—dest-karg-append <arg>

Add a kernel argument to each boot of the destination system.

—dest-karg-delete <arg>

Delete a kernel argument from each boot of the destination system.

—network-keyfile <path>

Configure networking by using the specified NetworkManager keyfile for live and destination systems.

—ignition-ca <path>

Specify an additional TLS certificate authority to be trusted by Ignition.

—pre-install <path>

Run the specified script before installation.

—post-install <path>

Run the specified script after installation.

—installer-config <path>

Apply the specified installer configuration file.

—live-ignition <path>

Merge the specified Ignition config file into a new configuration fragment for the live environment.

—live-karg-append <arg>

Add a kernel argument to each boot of the live environment.

—live-karg-delete <arg>

Delete a kernel argument from each boot of the live environment.

—live-karg-replace <k=o=n>

Replace a kernel argument in each boot of the live environment, in the form key=old=new.

-f, —force

Overwrite an existing Ignition config.

-o, —output <path>

Write the ISO to a new output file.

-h, —help

Print help information.

coreos-installer PXE subcommands

Subcommand

Description

Note that not all of these options are accepted by all subcommands.

coreos-installer pxe customize <options> <path>

Customize a FCOS live PXE boot config.

coreos-installer pxe ignition wrap <options>

Wrap an Ignition config in an image.

coreos-installer pxe ignition unwrap <options> <image_name>

Show the wrapped Ignition config in an image.

coreos-installer PXE customize subcommand options

Option

Description

Note that not all of these options are accepted by all subcommands.

—dest-ignition <path>

Merge the specified Ignition config file into a new configuration fragment for the destination system.

—dest-device <path>

Install and overwrite the specified destination device.

—network-keyfile <path>

Configure networking by using the specified NetworkManager keyfile for live and destination systems.

—ignition-ca <path>

Specify an additional TLS certificate authority to be trusted by Ignition.

—pre-install <path>

Run the specified script before installation.

post-install <path>

Run the specified script after installation.

—installer-config <path>

Apply the specified installer configuration file.

—live-ignition <path>

Merge the specified Ignition config file into a new configuration fragment for the live environment.

-o, —output <path>

Write the initramfs to a new output file.

This option is required for PXE environments.

-h, —help

Print help information.

coreos.inst boot options for ISO or PXE installations

You can automatically invoke coreos-installer options at boot time by passing coreos.inst boot arguments to the FCOS live installer. These are provided in addition to the standard boot arguments.

  • For ISO installations, the coreos.inst options can be added by interrupting the automatic boot at the bootloader menu. You can interrupt the automatic boot by pressing TAB while the RHEL CoreOS (Live) menu option is highlighted.

  • For PXE or iPXE installations, the coreos.inst options must be added to the APPEND line before the FCOS live installer is booted.

The following table shows the FCOS live installer coreos.inst boot options for ISO and PXE installations.

Table 13. coreos.inst boot options
ArgumentDescription

coreos.inst.install_dev

Required. The block device on the system to install to. It is recommended to use the full path, such as /dev/sda, although sda is allowed.

coreos.inst.ignition_url

Optional: The URL of the Ignition config to embed into the installed system. If no URL is specified, no Ignition config is embedded. Only HTTP and HTTPS protocols are supported.

coreos.inst.save_partlabel

Optional: Comma-separated labels of partitions to preserve during the install. Glob-style wildcards are permitted. The specified partitions do not need to exist.

coreos.inst.save_partindex

Optional: Comma-separated indexes of partitions to preserve during the install. Ranges m-n are permitted, and either m or n can be omitted. The specified partitions do not need to exist.

coreos.inst.insecure

Optional: Permits the OS image that is specified by coreos.inst.image_url to be unsigned.

coreos.inst.image_url

Optional: Download and install the specified FCOS image.

  • This argument should not be used in production environments and is intended for debugging purposes only.

  • While this argument can be used to install a version of FCOS that does not match the live media, it is recommended that you instead use the media that matches the version you want to install.

  • If you are using coreos.inst.image_url, you must also use coreos.inst.insecure. This is because the bare-metal media are not GPG-signed for OKD.

  • Only HTTP and HTTPS protocols are supported.

coreos.inst.skip_reboot

Optional: The system will not reboot after installing. After the install finishes, you will receive a prompt that allows you to inspect what is happening during installation. This argument should not be used in production environments and is intended for debugging purposes only.

coreos.inst.platform_id

Optional: The Ignition platform ID of the platform the FCOS image is being installed on. Default is metal. This option determines whether or not to request an Ignition config from the cloud provider, such as VMware. For example: coreos.inst.platform_id=vmware.

ignition.config.url

Optional: The URL of the Ignition config for the live boot. For example, this can be used to customize how coreos-installer is invoked, or to run code before or after the installation. This is different from coreos.inst.ignition_url, which is the Ignition config for the installed system.

Enabling multipathing with kernel arguments on FCOS

FCOS supports multipathing on the primary disk, allowing stronger resilience to hardware failure to achieve higher host availability.

You can enable multipathing at installation time for nodes that were provisioned in OKD 4.8 or later. While post-installation support is available by activating multipathing via the machine config, enabling multipathing during installation is recommended.

In setups where any I/O to non-optimized paths results in I/O system errors, you must enable multipathing at installation time.

On IBM Z and LinuxONE, you can enable multipathing only if you configured your cluster for it during installation. For more information, see “Installing FCOS and starting the OKD bootstrap process” in Installing a cluster with z/VM on IBM Z and LinuxONE.

The following procedure enables multipath at installation time and appends kernel arguments to the coreos-installer install command so that the installed system itself will use multipath beginning from the first boot.

Prerequisites

  • You have a running OKD cluster that uses version 4.8 or later.

  • You are logged in to the cluster as a user with administrative privileges.

Procedure

  1. To enable multipath and start the multipathd daemon, run the following command:

    1. $ mpathconf --enable && systemctl start multipathd.service
    • Optional: If booting the PXE or ISO, you can instead enable multipath by adding rd.multipath=default from the kernel command line.
  2. Append the kernel arguments by invoking the coreos-installer program:

    • If there is only one multipath device connected to the machine, it should be available at path /dev/mapper/mpatha. For example:

      1. $ coreos-installer install /dev/mapper/mpatha \ (1)
      2. --append-karg rd.multipath=default \
      3. --append-karg root=/dev/disk/by-label/dm-mpath-root \
      4. --append-karg rw
      1Indicates the path of the single multipathed device.
    • If there are multiple multipath devices connected to the machine, or to be more explicit, instead of using /dev/mapper/mpatha, it is recommended to use the World Wide Name (WWN) symlink available in /dev/disk/by-id. For example:

      1. $ coreos-installer install /dev/disk/by-id/wwn-<wwn_ID> \ (1)
      2. --append-karg rd.multipath=default \
      3. --append-karg root=/dev/disk/by-label/dm-mpath-root \
      4. --append-karg rw
      1Indicates the WWN ID of the target multipathed device. For example, 0xx194e957fcedb4841.

      This symlink can also be used as the coreos.inst.install_dev kernel argument when using special coreos.inst.* arguments to direct the live installer. For more information, see “Installing FCOS and starting the OKD bootstrap process”.

  3. Check that the kernel arguments worked by going to one of the worker nodes and listing the kernel command line arguments (in /proc/cmdline on the host):

    1. $ oc debug node/ip-10-0-141-105.ec2.internal

    Example output

    1. Starting pod/ip-10-0-141-105ec2internal-debug ...
    2. To use host binaries, run `chroot /host`
    3. sh-4.2# cat /host/proc/cmdline
    4. ...
    5. rd.multipath=default root=/dev/disk/by-label/dm-mpath-root
    6. ...
    7. sh-4.2# exit

    You should see the added kernel arguments.

Updating the bootloader using bootupd

To update the bootloader by using bootupd, you must either install bootupd on FCOS machines manually or provide a machine config with the enabled systemd unit. Unlike grubby or other bootloader tools, bootupd does not manage kernel space configuration such as passing kernel arguments.

After you have installed bootupd, you can manage it remotely from the OKD cluster.

It is recommended that you use bootupd only on bare metal or virtualized hypervisor installations, such as for protection against the BootHole vulnerability.

Manual install method

You can manually install bootupd by using the bootctl command-line tool.

  1. Inspect the system status:

    1. # bootupctl status

    Example output for x86_64

    1. Component EFI
    2. Installed: grub2-efi-x64-1:2.04-31.fc33.x86_64,shim-x64-15-8.x86_64
    3. Update: At latest version

    Example output for arm64

    1. Component EFI
    2. Installed: grub2-efi-aa64-1:2.02-99.el8_4.1.aarch64,shim-aa64-15.4-2.el8_1.aarch64
    3. Update: At latest version
  2. FCOS images created without bootupd installed on them require an explicit adoption phase.

    If the system status is Adoptable, perform the adoption:

    1. # bootupctl adopt-and-update

    Example output

    1. Updated: grub2-efi-x64-1:2.04-31.fc33.x86_64,shim-x64-15-8.x86_64
  3. If an update is available, apply the update so that the changes take effect on the next reboot:

    1. # bootupctl update

    Example output

    1. Updated: grub2-efi-x64-1:2.04-31.fc33.x86_64,shim-x64-15-8.x86_64

Machine config method

Another way to enable bootupd is by providing a machine config.

  • Provide a machine config file with the enabled systemd unit, as shown in the following example:

    Example output

    1. variant: rhcos
    2. version: 1.1.0
    3. systemd:
    4. units:
    5. - name: custom-bootupd-auto.service
    6. enabled: true
    7. contents: |
    8. [Unit]
    9. Description=Bootupd automatic update
    10. [Service]
    11. ExecStart=/usr/bin/bootupctl update
    12. RemainAfterExit=yes
    13. [Install]
    14. WantedBy=multi-user.target

Waiting for the bootstrap process to complete

The OKD bootstrap process begins after the cluster nodes first boot into the persistent FCOS environment that has been installed to disk. The configuration information provided through the Ignition config files is used to initialize the bootstrap process and install OKD on the machines. You must wait for the bootstrap process to complete.

Prerequisites

  • You have created the Ignition config files for your cluster.

  • You have configured suitable network, DNS and load balancing infrastructure.

  • You have obtained the installation program and generated the Ignition config files for your cluster.

  • You installed FCOS on your cluster machines and provided the Ignition config files that the OKD installation program generated.

Procedure

  1. Monitor the bootstrap process:

    1. $ ./openshift-install --dir <installation_directory> wait-for bootstrap-complete \ (1)
    2. --log-level=info (2)
    1For <installation_directory>, specify the path to the directory that you stored the installation files in.
    2To view different installation details, specify warn, debug, or error instead of info.

    Example output

    1. INFO Waiting up to 30m0s for the Kubernetes API at https://api.test.example.com:6443...
    2. INFO API v1.23.0 up
    3. INFO Waiting up to 30m0s for bootstrapping to complete...
    4. INFO It is now safe to remove the bootstrap resources

    The command succeeds when the Kubernetes API server signals that it has been bootstrapped on the control plane machines.

  2. After bootstrap process is complete, remove the bootstrap machine from the load balancer.

    You must remove the bootstrap machine from the load balancer at this point. You can also remove or reformat the bootstrap machine itself.

Additional resources

Logging in to the cluster by using the CLI

You can log in to your cluster as a default system user by exporting the cluster kubeconfig file. The kubeconfig file contains information about the cluster that is used by the CLI to connect a client to the correct cluster and API server. The file is specific to a cluster and is created during OKD installation.

Prerequisites

  • You deployed an OKD cluster.

  • You installed the oc CLI.

Procedure

  1. Export the kubeadmin credentials:

    1. $ export KUBECONFIG=<installation_directory>/auth/kubeconfig (1)
    1For <installation_directory>, specify the path to the directory that you stored the installation files in.
  2. Verify you can run oc commands successfully using the exported configuration:

    1. $ oc whoami

    Example output

    1. system:admin

Approving the certificate signing requests for your machines

When you add machines to a cluster, two pending certificate signing requests (CSRs) are generated for each machine that you added. You must confirm that these CSRs are approved or, if necessary, approve them yourself. The client requests must be approved first, followed by the server requests.

Prerequisites

  • You added machines to your cluster.

Procedure

  1. Confirm that the cluster recognizes the machines:

    1. $ oc get nodes

    Example output

    1. NAME STATUS ROLES AGE VERSION
    2. master-0 Ready master 63m v1.23.0
    3. master-1 Ready master 63m v1.23.0
    4. master-2 Ready master 64m v1.23.0

    The output lists all of the machines that you created.

    The preceding output might not include the compute nodes, also known as worker nodes, until some CSRs are approved.

  2. Review the pending CSRs and ensure that you see the client requests with the Pending or Approved status for each machine that you added to the cluster:

    1. $ oc get csr

    Example output

    1. NAME AGE REQUESTOR CONDITION
    2. csr-8b2br 15m system:serviceaccount:openshift-machine-config-operator:node-bootstrapper Pending
    3. csr-8vnps 15m system:serviceaccount:openshift-machine-config-operator:node-bootstrapper Pending
    4. ...

    In this example, two machines are joining the cluster. You might see more approved CSRs in the list.

  3. If the CSRs were not approved, after all of the pending CSRs for the machines you added are in Pending status, approve the CSRs for your cluster machines:

    Because the CSRs rotate automatically, approve your CSRs within an hour of adding the machines to the cluster. If you do not approve them within an hour, the certificates will rotate, and more than two certificates will be present for each node. You must approve all of these certificates. After the client CSR is approved, the Kubelet creates a secondary CSR for the serving certificate, which requires manual approval. Then, subsequent serving certificate renewal requests are automatically approved by the machine-approver if the Kubelet requests a new certificate with identical parameters.

    For clusters running on platforms that are not machine API enabled, such as bare metal and other user-provisioned infrastructure, you must implement a method of automatically approving the kubelet serving certificate requests (CSRs). If a request is not approved, then the oc exec, oc rsh, and oc logs commands cannot succeed, because a serving certificate is required when the API server connects to the kubelet. Any operation that contacts the Kubelet endpoint requires this certificate approval to be in place. The method must watch for new CSRs, confirm that the CSR was submitted by the node-bootstrapper service account in the system:node or system:admin groups, and confirm the identity of the node.

    • To approve them individually, run the following command for each valid CSR:

      1. $ oc adm certificate approve <csr_name> (1)
      1<csr_name> is the name of a CSR from the list of current CSRs.
    • To approve all pending CSRs, run the following command:

      1. $ oc get csr -o go-template='{{range .items}}{{if not .status}}{{.metadata.name}}{{"\n"}}{{end}}{{end}}' | xargs --no-run-if-empty oc adm certificate approve

      Some Operators might not become available until some CSRs are approved.

  4. Now that your client requests are approved, you must review the server requests for each machine that you added to the cluster:

    1. $ oc get csr

    Example output

    1. NAME AGE REQUESTOR CONDITION
    2. csr-bfd72 5m26s system:node:ip-10-0-50-126.us-east-2.compute.internal Pending
    3. csr-c57lv 5m26s system:node:ip-10-0-95-157.us-east-2.compute.internal Pending
    4. ...
  5. If the remaining CSRs are not approved, and are in the Pending status, approve the CSRs for your cluster machines:

    • To approve them individually, run the following command for each valid CSR:

      1. $ oc adm certificate approve <csr_name> (1)
      1<csr_name> is the name of a CSR from the list of current CSRs.
    • To approve all pending CSRs, run the following command:

      1. $ oc get csr -o go-template='{{range .items}}{{if not .status}}{{.metadata.name}}{{"\n"}}{{end}}{{end}}' | xargs oc adm certificate approve
  6. After all client and server CSRs have been approved, the machines have the Ready status. Verify this by running the following command:

    1. $ oc get nodes

    Example output

    1. NAME STATUS ROLES AGE VERSION
    2. master-0 Ready master 73m v1.23.0
    3. master-1 Ready master 73m v1.23.0
    4. master-2 Ready master 74m v1.23.0
    5. worker-0 Ready worker 11m v1.23.0
    6. worker-1 Ready worker 11m v1.23.0

    It can take a few minutes after approval of the server CSRs for the machines to transition to the Ready status.

Additional information

Initial Operator configuration

After the control plane initializes, you must immediately configure some Operators so that they all become available.

Prerequisites

  • Your control plane has initialized.

Procedure

  1. Watch the cluster components come online:

    1. $ watch -n5 oc get clusteroperators

    Example output

    1. NAME VERSION AVAILABLE PROGRESSING DEGRADED SINCE
    2. authentication 4.10.0 True False False 19m
    3. baremetal 4.10.0 True False False 37m
    4. cloud-credential 4.10.0 True False False 40m
    5. cluster-autoscaler 4.10.0 True False False 37m
    6. config-operator 4.10.0 True False False 38m
    7. console 4.10.0 True False False 26m
    8. csi-snapshot-controller 4.10.0 True False False 37m
    9. dns 4.10.0 True False False 37m
    10. etcd 4.10.0 True False False 36m
    11. image-registry 4.10.0 True False False 31m
    12. ingress 4.10.0 True False False 30m
    13. insights 4.10.0 True False False 31m
    14. kube-apiserver 4.10.0 True False False 26m
    15. kube-controller-manager 4.10.0 True False False 36m
    16. kube-scheduler 4.10.0 True False False 36m
    17. kube-storage-version-migrator 4.10.0 True False False 37m
    18. machine-api 4.10.0 True False False 29m
    19. machine-approver 4.10.0 True False False 37m
    20. machine-config 4.10.0 True False False 36m
    21. marketplace 4.10.0 True False False 37m
    22. monitoring 4.10.0 True False False 29m
    23. network 4.10.0 True False False 38m
    24. node-tuning 4.10.0 True False False 37m
    25. openshift-apiserver 4.10.0 True False False 32m
    26. openshift-controller-manager 4.10.0 True False False 30m
    27. openshift-samples 4.10.0 True False False 32m
    28. operator-lifecycle-manager 4.10.0 True False False 37m
    29. operator-lifecycle-manager-catalog 4.10.0 True False False 37m
    30. operator-lifecycle-manager-packageserver 4.10.0 True False False 32m
    31. service-ca 4.10.0 True False False 38m
    32. storage 4.10.0 True False False 37m
  2. Configure the Operators that are not available.

Additional resources

Disabling the default OperatorHub sources

Operator catalogs that source content provided by Red Hat and community projects are configured for OperatorHub by default during an OKD installation. In a restricted network environment, you must disable the default catalogs as a cluster administrator.

Procedure

  • Disable the sources for the default catalogs by adding disableAllDefaultSources: true to the OperatorHub object:

    1. $ oc patch OperatorHub cluster --type json \
    2. -p '[{"op": "add", "path": "/spec/disableAllDefaultSources", "value": true}]'

Alternatively, you can use the web console to manage catalog sources. From the AdministrationCluster SettingsConfigurationOperatorHub page, click the Sources tab, where you can create, delete, disable, and enable individual sources.

Image registry storage configuration

The Image Registry Operator is not initially available for platforms that do not provide default storage. After installation, you must configure your registry to use storage so that the Registry Operator is made available.

Instructions are shown for configuring a persistent volume, which is required for production clusters. Where applicable, instructions are shown for configuring an empty directory as the storage location, which is available for only non-production clusters.

Additional instructions are provided for allowing the image registry to use block storage types by using the Recreate rollout strategy during upgrades.

Changing the image registry’s management state

To start the image registry, you must change the Image Registry Operator configuration’s managementState from Removed to Managed.

Procedure

  • Change managementState Image Registry Operator configuration from Removed to Managed. For example:

    1. $ oc patch configs.imageregistry.operator.openshift.io cluster --type merge --patch '{"spec":{"managementState":"Managed"}}'

Configuring registry storage for bare metal and other manual installations

As a cluster administrator, following installation you must configure your registry to use storage.

Prerequisites

  • You have access to the cluster as a user with the cluster-admin role.

  • You have a cluster that uses manually-provisioned Fedora CoreOS (FCOS) nodes, such as bare metal.

  • You have provisioned persistent storage for your cluster, such as Red Hat OpenShift Data Foundation.

    OKD supports ReadWriteOnce access for image registry storage when you have only one replica. To deploy an image registry that supports high availability with two or more replicas, ReadWriteMany access is required.

  • Must have 100Gi capacity.

Procedure

  1. To configure your registry to use storage, change the spec.storage.pvc in the configs.imageregistry/cluster resource.

    When using shared storage, review your security settings to prevent outside access.

  2. Verify that you do not have a registry pod:

    1. $ oc get pod -n openshift-image-registry

    If the storage type is emptyDIR, the replica number cannot be greater than 1.

  3. Check the registry configuration:

    1. $ oc edit configs.imageregistry.operator.openshift.io

    Example output

    1. storage:
    2. pvc:
    3. claim:

    Leave the claim field blank to allow the automatic creation of an image-registry-storage PVC.

  4. Check the clusteroperator status:

    1. $ oc get clusteroperator image-registry
  5. Ensure that your registry is set to managed to enable building and pushing of images.

    • Run:

      1. $ oc edit configs.imageregistry/cluster

      Then, change the line

      1. managementState: Removed

      to

      1. managementState: Managed

Configuring storage for the image registry in non-production clusters

You must configure storage for the Image Registry Operator. For non-production clusters, you can set the image registry to an empty directory. If you do so, all images are lost if you restart the registry.

Procedure

  • To set the image registry storage to an empty directory:

    1. $ oc patch configs.imageregistry.operator.openshift.io cluster --type merge --patch '{"spec":{"storage":{"emptyDir":{}}}}'

    Configure this option for only non-production clusters.

    If you run this command before the Image Registry Operator initializes its components, the oc patch command fails with the following error:

    1. Error from server (NotFound): configs.imageregistry.operator.openshift.io "cluster" not found

    Wait a few minutes and run the command again.

Configuring block registry storage

To allow the image registry to use block storage types during upgrades as a cluster administrator, you can use the Recreate rollout strategy.

Block storage volumes are supported but not recommended for use with the image registry on production clusters. An installation where the registry is configured on block storage is not highly available because the registry cannot have more than one replica.

Procedure

  1. To set the image registry storage as a block storage type, patch the registry so that it uses the Recreate rollout strategy and runs with only one (1) replica:

    1. $ oc patch config.imageregistry.operator.openshift.io/cluster --type=merge -p '{"spec":{"rolloutStrategy":"Recreate","replicas":1}}'
  2. Provision the PV for the block storage device, and create a PVC for that volume. The requested block volume uses the ReadWriteOnce (RWO) access mode.

  3. Edit the registry configuration so that it references the correct PVC.

Completing installation on user-provisioned infrastructure

After you complete the Operator configuration, you can finish installing the cluster on infrastructure that you provide.

Prerequisites

  • Your control plane has initialized.

  • You have completed the initial Operator configuration.

Procedure

  1. Confirm that all the cluster components are online with the following command:

    1. $ watch -n5 oc get clusteroperators

    Example output

    1. NAME VERSION AVAILABLE PROGRESSING DEGRADED SINCE
    2. authentication 4.10.0 True False False 19m
    3. baremetal 4.10.0 True False False 37m
    4. cloud-credential 4.10.0 True False False 40m
    5. cluster-autoscaler 4.10.0 True False False 37m
    6. config-operator 4.10.0 True False False 38m
    7. console 4.10.0 True False False 26m
    8. csi-snapshot-controller 4.10.0 True False False 37m
    9. dns 4.10.0 True False False 37m
    10. etcd 4.10.0 True False False 36m
    11. image-registry 4.10.0 True False False 31m
    12. ingress 4.10.0 True False False 30m
    13. insights 4.10.0 True False False 31m
    14. kube-apiserver 4.10.0 True False False 26m
    15. kube-controller-manager 4.10.0 True False False 36m
    16. kube-scheduler 4.10.0 True False False 36m
    17. kube-storage-version-migrator 4.10.0 True False False 37m
    18. machine-api 4.10.0 True False False 29m
    19. machine-approver 4.10.0 True False False 37m
    20. machine-config 4.10.0 True False False 36m
    21. marketplace 4.10.0 True False False 37m
    22. monitoring 4.10.0 True False False 29m
    23. network 4.10.0 True False False 38m
    24. node-tuning 4.10.0 True False False 37m
    25. openshift-apiserver 4.10.0 True False False 32m
    26. openshift-controller-manager 4.10.0 True False False 30m
    27. openshift-samples 4.10.0 True False False 32m
    28. operator-lifecycle-manager 4.10.0 True False False 37m
    29. operator-lifecycle-manager-catalog 4.10.0 True False False 37m
    30. operator-lifecycle-manager-packageserver 4.10.0 True False False 32m
    31. service-ca 4.10.0 True False False 38m
    32. storage 4.10.0 True False False 37m

    Alternatively, the following command notifies you when all of the clusters are available. It also retrieves and displays credentials:

    1. $ ./openshift-install --dir <installation_directory> wait-for install-complete (1)
    1For <installation_directory>, specify the path to the directory that you stored the installation files in.

    Example output

    1. INFO Waiting up to 30m0s for the cluster to initialize...

    The command succeeds when the Cluster Version Operator finishes deploying the OKD cluster from Kubernetes API server.

    • The Ignition config files that the installation program generates contain certificates that expire after 24 hours, which are then renewed at that time. If the cluster is shut down before renewing the certificates and the cluster is later restarted after the 24 hours have elapsed, the cluster automatically recovers the expired certificates. The exception is that you must manually approve the pending node-bootstrapper certificate signing requests (CSRs) to recover kubelet certificates. See the documentation for Recovering from expired control plane certificates for more information.

    • It is recommended that you use Ignition config files within 12 hours after they are generated because the 24-hour certificate rotates from 16 to 22 hours after the cluster is installed. By using the Ignition config files within 12 hours, you can avoid installation failure if the certificate update runs during installation.

  2. Confirm that the Kubernetes API server is communicating with the pods.

    1. To view a list of all pods, use the following command:

      1. $ oc get pods --all-namespaces

      Example output

      1. NAMESPACE NAME READY STATUS RESTARTS AGE
      2. openshift-apiserver-operator openshift-apiserver-operator-85cb746d55-zqhs8 1/1 Running 1 9m
      3. openshift-apiserver apiserver-67b9g 1/1 Running 0 3m
      4. openshift-apiserver apiserver-ljcmx 1/1 Running 0 1m
      5. openshift-apiserver apiserver-z25h4 1/1 Running 0 2m
      6. openshift-authentication-operator authentication-operator-69d5d8bf84-vh2n8 1/1 Running 0 5m
      7. ...
    2. View the logs for a pod that is listed in the output of the previous command by using the following command:

      1. $ oc logs <pod_name> -n <namespace> (1)
      1Specify the pod name and namespace, as shown in the output of the previous command.

      If the pod logs display, the Kubernetes API server can communicate with the cluster machines.

  3. For an installation with Fibre Channel Protocol (FCP), additional steps are required to enable multipathing. Do not enable multipathing during installation.

    See “Enabling multipathing with kernel arguments on FCOS” in the Post-installation machine configuration tasks documentation for more information.

  4. Register your cluster on the Cluster registration page.

Additional resources

Next steps