Chapter 2. Installing

The following apply to production environments. Test or sample environments will function with the minimum requirements.

When planning an environment with multiple masters, a minimum of three etcd hosts as well as a load-balancer between the master hosts, is required.

Use the above with the following table to plan the maximum loads for nodes and pods:

By default, OpenShift Enterprise masters and nodes use all available cores in the system they run on. You can choose the number of cores you want OpenShift Enterprise to use by setting the GOMAXPROCS environment variable.

For example, run the following before starting the server to make OpenShift Enterprise only run on one core:

# export GOMAXPROCS=1

OpenShift Enterprise runs containers on your hosts, and in some cases, such as build operations and the registry service, it does so using privileged containers. Furthermore, those containers access your host’s Docker daemon and perform docker build and docker push operations. As such, you should be aware of the inherent security risks associated with performing docker run operations on arbitrary images as they effectively have root access.

For more information, see these articles:

To address these risks, OpenShift Enterprise uses security context constraints that control the actions that pods can perform and what it has the ability to access.

A fully functional DNS environment is a requirement for OpenShift Enterprise to work correctly. Adding entries into the /etc/hosts file is not enough, because that file is not copied into containers running on the platform.

To configure the OpenShift Enterprise DNS environment:

Key components of OpenShift Enterprise run themselves inside of containers. By default, these containers receive their /etc/resolv.conf DNS configuration file from their host. OpenShift Enterprise then inserts one DNS value into the pods (above the node’s nameserver values). That value is defined in the /etc/origin/node/node-config.yaml file by the dnsIP parameter, which by default is set to the address of the host node because the host is using dnsmasq. If the dnsIP parameter is omitted from the node-config.yaml file, then the value defaults to the kubernetes service IP, which is the first nameserver in the pod’s /etc/resolv.conf file.

As of OpenShift Enterprise 3.2, dnsmasq is automatically configured on all masters and nodes. The pods use the nodes as their DNS, and the nodes forward the requests. By default, dnsmasq is configured on the nodes to listen on port 53, therefore the nodes cannot run any other type of DNS application.

If you do not have a properly functioning DNS environment, you could experience failure with:

Configuring a DNS Environment

To properly configure your DNS environment for OpenShift Enterprise:

*.cloudapps.example.com. 300 IN  A 192.168.133.2

$ host `hostname`
ose3-master.example.com has address 172.16.25.41

$ dig ose3-node1.example.com  +short
172.16.25.45

A shared network must exist between the master and node hosts. If you plan to configure multiple masters for high-availability using the advanced installation method, you must also select an IP to be configured as your virtual IP (VIP) during the installation process. The IP that you select must be routable between all of your nodes, and if you configure using a FQDN it should resolve on all nodes.

NetworkManager

NetworkManager, a program for providing detection and configuration for systems to automatically connect to the network, is required.

Required Ports

OpenShift Enterprise infrastructure components communicate with each other using ports, which are communication endpoints that are identifiable for specific processes or services. Ensure the following ports required by OpenShift Enterprise are open between hosts, for example if you have a firewall in your environment. Some ports are optional depending on your configuration and usage.

Notes

You must have either Internet access and a GitHub account, or read and write access to an internal, HTTP-based Git server

The Kubernetes persistent volume framework allows you to provision an OpenShift Enterprise cluster with persistent storage using networked storage available in your environment. This can be done after completing the initial OpenShift Enterprise installation depending on your application needs, giving users a way to request those resources without having any knowledge of the underlying infrastructure.

The Installation and Configuration Guide provides instructions for cluster administrators on provisioning an OpenShift Enterprise cluster with persistent storage using NFS, GlusterFS, Ceph RBD, OpenStack Cinder, AWS Elastic Block Store (EBS), GCE Persistent Disks, and iSCSI.

Security-Enhanced Linux (SELinux) must be enabled on all of the servers before installing OpenShift Enterprise or the installer will fail. Also, configure SELINUXTYPE=targeted in the /etc/selinux/config file:

# This file controls the state of SELinux on the system.
# SELINUX= can take one of these three values:
#     enforcing - SELinux security policy is enforced.
#     permissive - SELinux prints warnings instead of enforcing.
#     disabled - No SELinux policy is loaded.
SELINUX=enforcing
# SELINUXTYPE= can take one of these three values:
#     targeted - Targeted processes are protected,
#     minimum - Modification of targeted policy. Only selected processes are protected.
#     mls - Multi Level Security protection.
SELINUXTYPE=targeted

Set up the Security Group

When installing on AWS or OpenStack, ensure that you set up the appropriate security groups. These are some ports that you should have in your security groups, without which the installation will fail. You may need more depending on the cluster configuration you want to install. For more information and to adjust your security groups accordingly, see Required Ports for more information.

If configuring ELBs for load balancing the masters and/or routers, you also need to configure Ingress and Egress security groups for the ELBs appropriately.

Override Detected IP Addresses and Host Names

Some deployments require that the user override the detected host names and IP addresses for the hosts. To see the default values, run the openshift_facts playbook:

# ansible-playbook playbooks/byo/openshift_facts.yml

Now, verify the detected common settings. If they are not what you expect them to be, you can override them.

The Advanced Installation topic discusses the available Ansible variables in greater detail.

In AWS, situations that require overriding the variables include:

If setting openshift_hostname to something other than the metadata-provided private-dns-name value, the native cloud integration for those providers will no longer work.

For EC2 hosts in particular, they must be deployed in a VPC that has both DNS host names and DNS resolution enabled, and openshift_hostname should not be overridden.

Post-Installation Configuration for Cloud Providers

Following the installation process, you can configure OpenShift Enterprise for AWS, OpenStack, or GCE.

Installing an Operating System

A base installation of RHEL 7.1 or later or RHEL Atomic Host 7.2.4 or later is required for master and node hosts. See the following documentation for the respective installation instructions, if required:

Registering the Hosts

Each host must be registered using Red Hat Subscription Manager (RHSM) and have an active OpenShift Enterprise subscription attached to access the required packages.

Managing Packages

For RHEL 7 systems:

For RHEL Atomic Host 7 systems:

Installing Docker

At this point, you should install Docker on all master and node hosts. This allows you to configure your Docker storage options before installing OpenShift Enterprise.

Docker containers and the images they are created from are stored in Docker’s storage back end. This storage is ephemeral and separate from any persistent storage allocated to meet the needs of your applications.

For RHEL Atomic Host

The default storage back end for Docker on RHEL Atomic Host is a thin pool logical volume, which is supported for production environments. You must ensure that enough space is allocated for this volume per the Docker storage requirements mentioned in System Requirements.

If you do not have enough allocated, see Managing Storage with Docker Formatted Containers for details on using docker-storage-setup and basic instructions on storage management in RHEL Atomic Host.

For RHEL

The default storage back end for Docker on RHEL 7 is a thin pool on loopback devices, which is not supported for production use and only appropriate for proof of concept environments. For production environments, you must create a thin pool logical volume and re-configure Docker to use that volume.

You can use the docker-storage-setup script included with Docker to create a thin pool device and configure Docker’s storage driver. This can be done after installing Docker and should be done before creating images or containers. The script reads configuration options from the /etc/sysconfig/docker-storage-setup file and supports three options for creating the logical volume:

Option A is the most robust option, however it requires adding an additional block device to your host before configuring Docker storage. Options B and C both require leaving free space available when provisioning your host.

Reconfiguring Docker Storage

Should you need to reconfigure Docker storage after having created the docker-pool, you should first remove the docker-pool logical volume. If you are using a dedicated volume group, you should also remove the volume group and any associated physical volumes before reconfiguring docker-storage-setup according to the instructions above.

See Logical Volume Manager Administration for more detailed information on LVM management.

Managing Docker Container Logs

Sometimes a container’s log file (the /var/lib/docker/containers/<hash>/<hash>-json.log file on the node where the container is running) can increase to a problematic size. You can manage this by configuring Docker’s json-file logging driver to restrict the size and number of log files.

For example, to set the maximum file size to 1MB and always keep the last three log files, edit the /etc/sysconfig/docker file to configure max-size=1M and max-file=3:

OPTIONS='--insecure-registry=172.30.0.0/16 --selinux-enabled --log-opt max-size=1M --log-opt max-file=3'

Next, restart the Docker service:

# systemctl restart docker

Viewing Available Container Logs

Container logs are stored in the /var/lib/docker/containers/<hash>/ directory on the node where the container is running. For example:

# ls -lh /var/lib/docker/containers/f088349cceac173305d3e2c2e4790051799efe363842fdab5732f51f5b001fd8/
total 2.6M
-rw-r--r--. 1 root root 5.6K Nov 24 00:12 config.json
-rw-r--r--. 1 root root 649K Nov 24 00:15 f088349cceac173305d3e2c2e4790051799efe363842fdab5732f51f5b001fd8-json.log
-rw-r--r--. 1 root root 977K Nov 24 00:15 f088349cceac173305d3e2c2e4790051799efe363842fdab5732f51f5b001fd8-json.log.1
-rw-r--r--. 1 root root 977K Nov 24 00:15 f088349cceac173305d3e2c2e4790051799efe363842fdab5732f51f5b001fd8-json.log.2
-rw-r--r--. 1 root root 1.3K Nov 24 00:12 hostconfig.json
drwx------. 2 root root    6 Nov 24 00:12 secrets

See Docker’s documentation for additional information on how to Configure Logging Drivers.

The quick and advanced installation methods require a user that has access to all hosts. If you want to run the installer as a non-root user, passwordless sudo rights must be configured on each destination host.

For example, you can generate an SSH key on the host where you will invoke the installation process:

# ssh-keygen

Do not use a password.

An easy way to distribute your SSH keys is by using a bash loop:

# for host in master.example.com \
    node1.example.com \
    node2.example.com; \
    do ssh-copy-id -i ~/.ssh/id_rsa.pub $host; \
    done

Modify the host names in the above command according to your configuration.

To assign environment variables to hosts during the Ansible installation, indicate the desired variables in the /etc/ansible/hosts file after the host entry in the [masters] or [nodes] sections. For example:

[masters]
ec2-52-6-179-239.compute-1.amazonaws.com openshift_public_hostname=ose3-master.public.example.com

The following table describes variables for use with the Ansible installer that can be assigned to individual host entries:

To assign environment variables during the Ansible install that apply more globally to your OpenShift Enterprise cluster overall, indicate the desired variables in the /etc/ansible/hosts file on separate, single lines within the [OSEv3:vars] section. For example:

[OSEv3:vars]

openshift_master_identity_providers=[{'name': 'htpasswd_auth',
'login': 'true', 'challenge': 'true',
'kind': 'HTPasswdPasswordIdentityProvider',
'filename': '/etc/origin/master/htpasswd'}]

openshift_master_default_subdomain=apps.test.example.com

The following table describes variables for use with the Ansible installer that can be assigned cluster-wide:

If your hosts require use of a HTTP or HTTPS proxy in order to connect to external hosts, there are many components that must be configured to use the proxy, including masters, Docker, and builds. Node services only connect to the master API requiring no external access and therefore do not need to be configured to use a proxy.

In order to simplify this configuration, the following Ansible variables can be specified at a cluster or host level to apply these settings uniformly across your environment.

You can assign labels to node hosts during the Ansible install by configuring the /etc/ansible/hosts file. Labels are useful for determining the placement of pods onto nodes using the scheduler. Other than region=infra (discussed below), the actual label names and values are arbitrary and can be assigned however you see fit per your cluster’s requirements.

To assign labels to a node host during an Ansible install, use the openshift_node_labels variable with the desired labels added to the desired node host entry in the [nodes] section. In the following example, labels are set for a region called primary and a zone called east:

[nodes]
node1.example.com openshift_node_labels="{'region': 'primary', 'zone': 'east'}"

The openshift_router_selector and openshift_registry_selector Ansible settings are set to region=infra by default:

# default selectors for router and registry services
# openshift_router_selector='region=infra'
# openshift_registry_selector='region=infra'

The default router and registry will be automatically deployed if nodes exist that match the selector settings above. For example:

[nodes]
node1.example.com openshift_node_labels="{'region':'infra','zone':'default'}"

Any hosts you designate as masters during the installation process should also be configured as nodes by adding them to the [nodes] section so that the masters are configured as part of the OpenShift Enterprise SDN.

However, in order to ensure that your masters are not burdened with running pods, you can make them unschedulable by adding the openshift_schedulable=false option any node that is also a master. For example:

[nodes]
master.example.com openshift_node_labels="{'region':'infra','zone':'default'}" openshift_schedulable=false

Session options in the OAuth configuration are configurable in the inventory file. By default, Ansible populates a sessionSecretsFile with generated authentication and encryption secrets so that sessions generated by one master can be decoded by the others. The default location is /etc/origin/master/session-secrets.yaml, and this file will only be re-created if deleted on all masters.

You can set the session name and maximum number of seconds with openshift_master_session_name and openshift_master_session_max_seconds:

openshift_master_session_name=ssn
openshift_master_session_max_seconds=3600

If provided, openshift_master_session_auth_secrets and openshift_master_encryption_secrets must be equal length.

For openshift_master_session_auth_secrets, used to authenticate sessions using HMAC, it is recommended to use secrets with 32 or 64 bytes:

openshift_master_session_auth_secrets=['DONT+USE+THIS+SECRET+b4NV+pmZNSO']

For openshift_master_encryption_secrets, used to encrypt sessions, secrets must be 16, 24, or 32 characters long, to select AES-128, AES-192, or AES-256:

openshift_master_session_encryption_secrets=['DONT+USE+THIS+SECRET+b4NV+pmZNSO']

Custom serving certificates for the public host names of the OpenShift Enterprise API and web console can be deployed during an advanced installation and are configurable in the inventory file.

Certificate and key file paths can be configured using the openshift_master_named_certificates cluster variable:

openshift_master_named_certificates=[{"certfile": "/path/to/custom1.crt", "keyfile": "/path/to/custom1.key"}]

File paths must be local to the system where Ansible will be run. Certificates are copied to master hosts and are deployed within the /etc/origin/master/named_certificates/ directory.

Ansible detects a certificate’s Common Name and Subject Alternative Names. Detected names can be overridden by providing the "names" key when setting openshift_master_named_certificates:

openshift_master_named_certificates=[{"certfile": "/path/to/custom1.crt", "keyfile": "/path/to/custom1.key", "names": ["public-master-host.com"]}]

Certificates configured using openshift_master_named_certificates are cached on masters, meaning that each additional Ansible run with a different set of certificates results in all previously deployed certificates remaining in place on master hosts and within the master configuration file.

If you would like openshift_master_named_certificates to be overwritten with the provided value (or no value), specify the openshift_master_overwrite_named_certificates cluster variable:

openshift_master_overwrite_named_certificates=true

For a more complete example, consider the following cluster variables in an inventory file:

openshift_master_cluster_method=native
openshift_master_cluster_hostname=lb.openshift.com
openshift_master_cluster_public_hostname=custom.openshift.com

To overwrite the certificates on a subsequent Ansible run, you could set the following:

openshift_master_named_certificates=[{"certfile": "/root/STAR.openshift.com.crt", "keyfile": "/root/STAR.openshift.com.key", "names": ["custom.openshift.com"]}]
openshift_master_overwrite_named_certificates=true

You can also uninstall node components from specific hosts using the uninstall.yml playbook while leaving the remaining hosts and cluster alone:

When the playbook completes, all OpenShift Enterprise content should be removed from any specified hosts.

Before you sync with the required repositories, you may need to import the appropriate GPG key:

# rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-redhat-release

If the key is not imported, the indicated package is deleted after syncing the repository.

To sync the required repositories:

To sync the container images:

Container images can be exported from a system by first saving them to a tarball and then transporting them:

At this point you can perform the initial creation of the hosts that will be part of the OpenShift Enterprise environment. It is recommended to use the latest version of RHEL 7 and to perform a minimal installation. You will also want to pay attention to the other OpenShift Enterprise-specific prerequisites.

Once the hosts are initially built, the repositories can be set up.

On all of the relevant systems that will need OpenShift Enterprise software components, create the required repository definitions. Place the following text in the /etc/yum.repos.d/ose.repo file, replacing <server_IP> with the IP or host name of the Apache server hosting the software repositories:

[rhel-7-server-rpms]
name=rhel-7-server-rpms
baseurl=http://<server_IP>/repos/rhel-7-server-rpms
enabled=1
gpgcheck=0
[rhel-7-server-extras-rpms]
name=rhel-7-server-extras-rpms
baseurl=http://<server_IP>/repos/rhel-7-server-extras-rpms
enabled=1
gpgcheck=0
[rhel-7-server-ose-3.2-rpms]
name=rhel-7-server-ose-3.2-rpms
baseurl=http://<server_IP>/repos/rhel-7-server-ose-3.2-rpms
enabled=1
gpgcheck=0

At this point, the systems are ready to continue to be prepared following the OpenShift Enterprise documentation.

Skip the section titled Registering the Hosts and start with Managing Packages.

To import the relevant components, securely copy the images from the connected host to the individual OpenShift Enterprise hosts:

# scp /var/www/html/repos/images/ose3-images.tar root@<openshift_host_name>:
# ssh root@<openshift_host_name> "docker load -i ose3-images.tar"

If you prefer, you could use wget on each OpenShift Enterprise host to fetch the tar file, and then perform the Docker import command locally. Perform the same steps for the metrics and logging images, if you synchronized them.

On the host that will act as an OpenShift Enterprise master, copy and import the builder images:

# scp /var/www/html/images/ose3-builder-images.tar root@<openshift_master_host_name>:
# ssh root@<openshift_master_host_name> "docker load -i ose3-builder-images.tar"

You can now choose to follow the quick or advanced OpenShift Enterprise installation instructions in the documentation.

You now need to create the internal Docker registry.

Pushing the container images into OpenShift Enterprise’s Docker registry requires a user with cluster-admin privileges. Because the default OpenShift Enterprise system administrator does not have a standard authorization token, they cannot be used to log in to the Docker registry.

To create an administrative user:

The openshift project is where all of the image streams for builder images are created by the installer. They are loaded by the installer from the /usr/share/openshift/examples directory. Change all of the definitions by deleting the image streams which had been loaded into OpenShift Enterprise’s database, then re-create them:

At this point the system is ready to load the container images.

By default, the registry is created with no settings for compute resource requests or limits. For production, it is highly recommended that the deployment configuration for the registry be updated to set resource requests and limits for the registry pod. Otherwise, the registry pod will be considered a BestEffort pod.

See Compute Resources for more information on configuring requests and limits.

The registry stores Docker images and metadata. If you simply deploy a pod with the registry, it uses an ephemeral volume that is destroyed if the pod exits. Any images anyone has built or pushed into the registry would disappear.

$ oc volume deploymentconfigs/docker-registry --add --name=registry-storage -t pvc \
     --claim-name=<pvc_name> --overwrite

$ oc volume deploymentconfigs/docker-registry \
     --add --overwrite --name=registry-storage --mount-path=/registry \
     --source='{"nfs": { "server": "<fqdn>", "path": "/path/to/export"}}'

storage:
  cache:
    layerinfo: inmemory
  delete:
    enabled: true
  s3:
    accesskey: awsaccesskey
    secretkey: awssecretkey
    region: us-west-1
    regionendpoint: http://myobjects.local
    bucket: bucketname
    encrypt: true
    keyid: mykeyid
    secure: true
    v4auth: false
    chunksize: 5242880
    rootdirectory: /s3/object/name/prefix

$ oadm registry --service-account=registry \
    --config=/etc/origin/master/admin.kubeconfig \
    --images='registry.access.redhat.com/openshift3/ose-${component}:${version}' \
    --mount-host=<path>

$ sudo chown 1001:root <path>

OpenShift Enterprise refers to the integrated registry by its service IP address, so if you decide to delete and recreate the docker-registry service, you can ensure a completely transparent transition by arranging to re-use the old IP address in the new service. If a new IP address cannot be avoided, you can minimize cluster disruption by rebooting only the masters.

To access the registry directly, the user that you use must satisfy the following, depending on your intended usage:

For more information on user permissions, see Managing Role Bindings.

To log in to the registry directly:

After logging in to the registry, you can perform docker pull and docker push operations against your registry.

In the following examples, we use:

To enable managing the registry configuration file directly, it is recommended that the configuration file be mounted as a secret volume:

This may be performed as an iterative process to achieve the desired configuration. For example, during troubleshooting, the configuration may be temporarily updated to put it in debug mode.

To update an existing configuration:

There are many configuration options available in the upstream docker distribution library. Not all configuration options are supported or enabled. Use this section as a reference.

log:
  level: debug
  formatter: text
  fields:
    service: registry
    environment: staging

storage:
  delete:
    enabled: true 1
  redirect:
    disable: false
  cache:
    blobdescriptor: inmemory
  maintenance:
    uploadpurging:
      enabled: true
      age: 168h
      interval: 24h
      dryrun: false
    readonly:
      enabled: false

auth:
  openshift:
    realm: openshift

middleware:
  repository:
    - name: openshift 1
      options:
        pullthrough: true 2

version: 0.1
log:
  level: debug
http:
  addr: :5000
storage:
  cache:
    blobdescriptor: inmemory
  delete:
    enabled: true
  s3: 1
    accesskey: BJKMSZBRESWJQXRWMAEQ
    secretkey: 5ah5I91SNXbeoUXXDasFtadRqOdy62JzlnOW1goS
    region: us-east-1
    bucket: docker.myregistry.com
auth:
  openshift:
    realm: openshift
middleware:
  registry:
    - name: openshift
  repository:
    - name: openshift
   storage:
    - name: cloudfront 2
      options:
        baseurl: https://jrpbyn0k5k88bi.cloudfront.net/ 3
        privatekey: /etc/docker/cloudfront-ABCEDFGHIJKLMNOPQRST.pem 4
        keypairid: ABCEDFGHIJKLMNOPQRST 5
    - name: openshift

middleware:
  repository:
    - name: openshift
      options:
        acceptschema2: false 1
        enforcequota: false 2
        projectcachettl: 1m 3
        blobrepositorycachettl: 10m 4

http:
  addr: :5000
  tls:
    certificate: /etc/secrets/registry.crt
    key: /etc/secrets/registry.key

notifications:
  endpoints:
    - name: registry
      disabled: false
      url: https://url:port/path
      headers:
        Accept:
          - text/plain
      timeout: 500
      threshold: 5
      backoff: 1000

When using a scaled registry with a shared NFS volume, you may see one of the following errors during the push of an image:

These errors are returned by an internal registry service when Docker attempts to push the image. Its cause originates in the synchronization of file attributes across nodes. Factors such as NFS client side caching, network latency, and layer size can all contribute to potential errors that might occur when pushing an image using the default round-robin load balancing configuration.

You can perform the following steps to minimize the probability of such a failure:

This error occurs when the pulled image is pushed to an image stream different from the one it is being pulled from. This is caused by re-tagging a built image into an arbitrary image stream:

$ oc tag srcimagestream:latest anyproject/pullimagestream:latest

And subsequently pulling from it, using an image reference such as:

internal.registry.url:5000/anyproject/pullimagestream:latest

During a manual Docker pull, this will produce a similar error:

Error: image anyproject/pullimagestream:latest not found

To prevent this, avoid the tagging of internally managed images completely, or re-push the built image to the desired namespace manually.

There are problems reported happening when the registry runs on S3 storage back-end. Pushing to a Docker registry occasionally fails with the following error:

Received unexpected HTTP status: 500 Internal Server Error

To debug this, you need to view the registry logs. In there, look for similar error messages occurring at the time of the failed push:

time="2016-03-30T15:01:21.22287816-04:00" level=error msg="unknown error completing upload: driver.Error{DriverName:\"s3\", Enclosed:(*url.Error)(0xc20901cea0)}" http.request.method=PUT
...
time="2016-03-30T15:01:21.493067808-04:00" level=error msg="response completed with error" err.code=UNKNOWN err.detail="s3: Put https://s3.amazonaws.com/oso-tsi-docker/registry/docker/registry/v2/blobs/sha256/ab/abe5af443833d60cf672e2ac57589410dddec060ed725d3e676f1865af63d2e2/data: EOF" err.message="unknown error" http.request.method=PUT
...
time="2016-04-02T07:01:46.056520049-04:00" level=error msg="error putting into main store: s3: The request signature we calculated does not match the signature you provided. Check your key and signing method." http.request.method=PUT
atest

If you see such errors, contact your Amazon S3 support. There may be a problem in your region or with your particular bucket.

Check your registry log. If you see similar error message to the one below:

time="2016-08-10T07:29:06.882023903Z" level=panic msg="Configuration error: OpenShift registry middleware not activated" 2016-08-10 07:29:06.882174 I | http: panic serving 10.131.0.1:34558: &{0xc820010680 map[] 2016-08-10 07:29:06.882023903 +0000 UTC panic Configuration error: OpenShift registry middleware not activated}

It means that your custom configuration file lacks mandatory entries in the middleware section. Add them, re-deploy the registry, and restart your builds.

If you encounter the following error when pruning images:

BLOB sha256:49638d540b2b62f3b01c388e9d8134c55493b1fa659ed84e97cb59b87a6b8e6c error deleting blob

And your registry log contains the following information:

error deleting blob \"sha256:49638d540b2b62f3b01c388e9d8134c55493b1fa659ed84e97cb59b87a6b8e6c\": operation unsupported

It means that your custom configuration file lacks mandatory entries in the storage section, namely storage:delete:enabled set to true. Add them, re-deploy the registry, and repeat your image pruning operation.

You can set up a highly-available router on your OpenShift Enterprise cluster using IP failover.

You can customize the service ports that a template router binds to by setting the environment variables ROUTER_SERVICE_HTTP_PORT and ROUTER_SERVICE_HTTPS_PORT. This can be done by creating a template router, then editing its deployment configuration.

The following example creates a router deployment with 0 replicas and customizes the router service HTTP and HTTPS ports, then scales it appropriately (to 1 replica).

$ oadm router --replicas=0 --ports='10080:10080,10443:10443' 1
$ oc set env dc/router ROUTER_SERVICE_HTTP_PORT=10080  \
                   ROUTER_SERVICE_HTTPS_PORT=10443
$ oc scale dc/router --replicas=1

The following is an example using iptables to open the custom router service ports.

$ iptables -A INPUT -p tcp --dport 10080 -j ACCEPT
$ iptables -A INPUT -p tcp --dport 10443 -j ACCEPT

An administrator can create multiple routers with the same definition to serve the same set of routes. By having different groups of routers with different namespace or route selectors, they can vary the routes that the router serves.

Multiple routers can be grouped to distribute routing load in the cluster and separate tenants to different routers or shards. Each router or shard in the group handles routes based on the selectors in the router. An administrator can create shards over the whole cluster using ROUTE_LABELS. A user can create shards over a namespace (project) by using NAMESPACE_LABELS.

Making specific routers deploy on specific nodes requires two steps:

The access controls are based on the service account that the router is run with.

Using NAMESPACE_LABELS and/or ROUTE_LABELS, a router can filter out the namespaces and/or routes that it should service. This enables you to partition routes amongst multiple router deployments effectively distributing the set of routes.

Example: A router deployment finops-router is run with route selector NAMESPACE_LABELS="name in (finance, ops)" and a router deployment dev-router is run with route selector NAMESPACE_LABELS="name=dev".

If all routes are in the 3 namespaces finance, ops or dev, then this could effectively distribute our routes across two router deployments.

In the above scenario, sharding becomes a special case of partitioning with no overlapping sets. Routes are divided amongst multiple router shards.

The criteria for route selection governs how the routes are distributed. It is possible to have routes that overlap accross multiple router deployments.

Example: In addition to the finops-router and dev-router in the example above, we also have an devops-router which is run with a route selector NAMESPACE_LABELS="name in (dev, ops)".

The routes in namespaces dev or ops now are serviced by two different router deployments. This becomes a case where we have partitioned the routes with an overlapping set.

In addition, this enables us to create more complex routing rules ala divert high priority traffic to the dedicated finops-router but send the lower priority ones to the devops-router.

NAMESPACE_LABELS allows filtering the projects to service and selecting all the routes from those projects. But we may want to partition routes based on other criteria in the routes themselves. The ROUTE_LABELS selector allows you to slice-and-dice the routes themselves.

Example: A router deployment prod-router is run with route selector ROUTE_LABELS="mydeployment=prod" and a router deployment devtest-router is run with route selector ROUTE_LABELS="mydeployment in (dev, test)"

Example assumes you have all the routes you wish to serviced tagged with a label "mydeployment=<tag>".

Router sharding lets you select how routes are distributed among a set of routers.

Router sharding is based on labels; you set labels on the routes in the pool, and express the desired subset of those routes for the router to serve with a selection expression via the oc set env command.

First, ensure that service account associated with the router has the cluster reader permission.

The rest of this section describes an extended example. Suppose there are 26 routes, named a — z, in the pool, with various labels:

sla=high       geo=east     hw=modest     dept=finance
sla=medium     geo=west     hw=strong     dept=dev
sla=low                                   dept=ops

These labels express the concepts: service level agreement, geographical location, hardware requirements, and department. The routes in the pool can have at most one label from each column. Some routes may have other labels, entirely, or none at all.

Here is a convenience script mkshard that ilustrates how oadm router, oc set env, and oc scale work together to make a router shard.

#!/bin/bash
# Usage: mkshard ID SELECTION-EXPRESSION
id=$1
sel="$2"
router=router-shard-$id           1
oadm router $router --replicas=0  2
dc=dc/router-shard-$id            3
oc set env   $dc ROUTE_LABELS="$sel"  4
oc scale $dc --replicas=3         5

Running mkshard several times creates several routers:

Because a router shard is a construct based on labels, you can modify either the labels (via oc label) or the selection expression.

This section extends the example started in the Creating Router Shards section, demonstrating how to change the selection expression.

Here is a convenience script modshard that modifies an existing router to use a new selection expression:

#!/bin/bash
# Usage: modshard ID SELECTION-EXPRESSION...
id=$1
shift
router=router-shard-$id       1
dc=dc/$router                 2
oc scale $dc --replicas=0     3
oc set env   $dc "$@"             4
oc scale $dc --replicas=3     5

For example, to expand the department for router-shard-3 to include ops as well as dev:

$ modshard 3 ROUTE_LABELS='dept in (dev, ops)'

The result is that router-shard-3 now selects routes g — s (the combined sets of g — k and l — s).

This example takes into account that there are only three departments in this example scenario, and specifies a department to leave out of the shard, thus achieving the same result as the preceding example:

$ modshard 3 ROUTE_LABELS='dept != finanace'

This example specifies shows three comma-separated qualities, and results in only route b being selected:

$ modshard 3 ROUTE_LABELS='hw=strong,type=dynamic,geo=west'

Similarly to ROUTE_LABELS, which involve a route’s labels, you can select routes based on the labels of the route’s namespace labels, with the NAMESPACE_LABELS environment variable. This example modifies router-shard-3 to serve routes whose namespace has the label frequency=weekly:

$ modshard 3 NAMESPACE_LABELS='frequency=weekly'

The last example combines ROUTE_LABELS and NAMESPACE_LABELS to select routes with label sla=low and whose namespace has the label frequency=weekly:

$ modshard 3 \
    NAMESPACE_LABELS='frequency=weekly' \
    ROUTE_LABELS='sla=low'

The routes for a project can be handled by a selected router by using NAMESPACE_LABELS. The router is given a selector for a NAMESPACE_LABELS label and the project that wants to use the router applies the NAMESPACE_LABELS label to its namespace.

First, ensure that service account associated with the router has the cluster reader permission. This permits the router to read the labels that are applied to the namespaces.

Now create and label the router:

$ oadm router ...  --service-account=router
$ oc set env dc/router NAMESPACE_LABELS="router=r1"

Because the router has a selector for a namespace, the router will handle routes for that namespace. So, for example:

$ oc label namespace default "router=r1"

Now create routes in the default namespace, and the route is available in the default router:

$ oc create -f route1.yaml

Now create a new project (namespace) and create a route, route2.

$ oc new-project p1
$ oc create -f route2.yaml

And notice the route is not available in your router. Now label namespace p1 with "router=r1"

$ oc label namespace p1 "router=r1"

Which makes the route available to the router.

Note that removing the label from the namespace won’t have immediate effect (as we don’t see the updates in the router), so if you redeploy/start a new router pod, you should see the unlabelled effects.

$ oc scale dc/router --replicas=0 && oc scale dc/router --replicas=1

You can customize the default routing subdomain by modifying the master configuration file. Routes that do not specify a host name would have one generated using this default routing subdomain.

routingConfig:
  subdomain: v3.openshift.test

no-route-hostname-mynamespace.v3.openshift.test

If an administrator wants to restrict all routes to a specific routing subdomain, they can pass the --force-subdomain option to the oadm router command. This forces the router to override any host names specified in a route and generate one based on the template provided to the --force-subdomain option.

The following example runs a router, which overrides the route host names using a custom subdomain template ${name}-${namespace}.apps.example.com.

$ oadm router --force-subdomain='${name}-${namespace}.apps.example.com'

A TLS-enabled route that does not include a certificate uses the router’s default certificate instead. In most cases, this certificate should be provided by a trusted certificate authority, but for convenience you can use the OpenShift Enterprise CA to create the certificate. For example:

$ CA=/etc/origin/master
$ oadm ca create-server-cert --signer-cert=$CA/ca.crt \
      --signer-key=$CA/ca.key --signer-serial=$CA/ca.serial.txt \
      --hostnames='*.cloudapps.example.com' \
      --cert=cloudapps.crt --key=cloudapps.key

The router expects the certificate and key to be in PEM format in a single file:

$ cat cloudapps.crt cloudapps.key $CA/ca.crt > cloudapps.router.pem

From there you can use the --default-cert flag:

$ oadm router --default-cert=cloudapps.router.pem --service-account=router \
    --credentials=${ROUTER_KUBECONFIG:-"$KUBECONFIG"}

Currently, password protected key files are not supported. HAProxy prompts for a password upon starting and does not have a way to automate this process. To remove a passphrase from a keyfile, you can run:

# openssl rsa -in <passwordProtectedKey.key> -out <new.key>

Here is an example of how to use a secure edge terminated route with TLS termination occurring on the router before traffic is proxied to the destination. The secure edge terminated route specifies the TLS certificate and key information. The TLS certificate is served by the router front end.

First, start up a router instance:

# oadm router --replicas=1 --service-account=router  \
    --credentials=${ROUTER_KUBECONFIG:-"$KUBECONFIG"}

Next, create a private key, csr and certificate for our edge secured route. The instructions on how to do that would be specific to your certificate authority and provider. For a simple self-signed certificate for a domain named www.example.test, see the example shown below:

# sudo openssl genrsa -out example-test.key 2048
#
# sudo openssl req -new -key example-test.key -out example-test.csr  \
  -subj "/C=US/ST=CA/L=Mountain View/O=OS3/OU=Eng/CN=www.example.test"
#
# sudo openssl x509 -req -days 366 -in example-test.csr  \
      -signkey example-test.key -out example-test.crt

Generate a route using the above certificate and key.

$ oc create route edge --service=my-service \
    --hostname=www.example.test \
    --key=example-test.key --cert=example-test.crt
route "my-service" created

Look at its definition.

$ oc get route/my-service -o yaml
apiVersion: v1
kind: Route
metadata:
  name:  my-service
spec:
  host: www.example.test
  to:
    kind: Service
    name: my-service
  tls:
    termination: edge
    key: |
      -----BEGIN PRIVATE KEY-----
      [...]
      -----END PRIVATE KEY-----
    certificate: |
      -----BEGIN CERTIFICATE-----
      [...]
      -----END CERTIFICATE-----

Make sure your DNS entry for www.example.test points to your router instance(s) and the route to your domain should be available. The example below uses curl along with a local resolver to simulate the DNS lookup:

# routerip="4.1.1.1"  #  replace with IP address of one of your router instances.
# curl -k --resolve www.example.test:443:$routerip https://www.example.test/

The OpenShift Enterprise router runs inside a container and the default behavior is to use the network stack of the host (i.e., the node where the router container runs). This default behavior benefits performance because network traffic from remote clients does not need to take multiple hops through user space to reach the target service and container.

Additionally, this default behavior enables the router to get the actual source IP address of the remote connection rather than getting the node’s IP address. This is useful for defining ingress rules based on the originating IP, supporting sticky sessions, and monitoring traffic, among other uses.

This host network behavior is controlled by the --host-network router command line option, and the default behaviour is the equivalent of using --host-network=true. If you wish to run the router with the container network stack, use the --host-network=false option when creating the router. For example:

$ oadm router \
    --credentials='/etc/origin/master/openshift-router.kubeconfig' \
    --service-account=router \
    --host-network=false

Internally, this means the router container must publish the 80 and 443 ports in order for the external network to communicate with the router.

Using the --metrics-image and --expose-metrics options, you can configure the OpenShift Enterprise router to run a sidecar container that exposes or publishes router metrics for consumption by external metrics collection and aggregation systems (e.g. Prometheus, statsd).

Depending on your router implementation, the image is appropriately set up and the metrics sidecar container is started when the router is deployed. For example, the HAProxy-based router implementation defaults to using the prom/haproxy-exporter image to run as a sidecar container, which can then be used as a metrics datasource by the Prometheus server.

If you connect to the router while the proxy is reloading, there is a small chance that your connection will end up in the wrong network queue and be dropped. The issue is being addressed. In the meantime, it is possible to work around the problem by installing iptables rules to prevent connections during the reload window. However, doing so means that the router needs to run with elevated privilege so that it can manipulate iptables on the host. It also means that connections that happen during the reload are temporarily ignored and must retransmit their connection start, lengthening the time it takes to connect, but preventing connection failure.

To prevent this, configure the router to use iptables by changing the service account, and setting an environment variable on the router.

Use a Privileged SCC

When creating the router, allow it to use the privileged SCC. This gives the router user the ability to create containers with root privileges on the nodes:

$ oadm policy add-scc-to-user privileged -z router

Patch the Router Deployment Configuration to Create a Privileged Container

You can now create privileged containers. Next, configure the router deployment configuration to use the privilege so that the router can set the iptables rules it needs. This patch changes the router deployment configuration so that the container that is created runs as root:

$ oc patch dc router -p '{"spec":{"template":{"spec":{"containers":[{"name":"router","securityContext":{"privileged":true}}]}}}}'

Configure the Router to Use iptables

Set the option on the router deployment configuration:

$ oc set env dc/router -c router DROP_SYN_DURING_RESTART=1

If you used a non-default name for the router, you must change dc/router accordingly.

You can use ConfigMap to customize the router instance without rebuilding the router image. The haproxy-config.template, reload-haproxy, and other scripts can be modified as well as creating and modifying router environment variables.

The following example customization can be used in a highly-available routing setup to use stick-tables that synchronize between peers.

Adding a Peer Section

In order to synchronize stick-tables amongst peers you must a define a peers section in your HAProxy configuration. This section determines how HAProxy will identify and connect to peers. The plug-in provides data to the template under the .PeerEndpoints variable to allow you to easily identify members of the router service. You may add a peer section to the haproxy-config.template file inside the router image by adding:

{{ if (len .PeerEndpoints) gt 0 }}
peers openshift_peers
  {{ range $endpointID, $endpoint := .PeerEndpoints }}
    peer {{$endpoint.TargetName}} {{$endpoint.IP}}:1937
  {{ end }}
{{ end }}

Changing the Reload Script

When using stick-tables, you have the option of telling HAProxy what it should consider the name of the local host in the peer section. When creating endpoints, the plug-in attempts to set the TargetName to the value of the endpoint’s TargetRef.Name. If TargetRef is not set, it will set the TargetName to the IP address. The TargetRef.Name corresponds with the Kubernetes host name, therefore you can add the -L option to the reload-haproxy script to identify the local host in the peer section.

peer_name=$HOSTNAME 1

if [ -n "$old_pid" ]; then
  /usr/sbin/haproxy -f $config_file -p $pid_file -L $peer_name -sf $old_pid
else
  /usr/sbin/haproxy -f $config_file -p $pid_file -L $peer_name
fi

Modifying Back Ends

Finally, to use the stick-tables within back ends, you can modify the HAProxy configuration to use the stick-tables and peer set. The following is an example of changing the existing back end for TCP connections to use stick-tables:

            {{ if eq $cfg.TLSTermination "passthrough" }}
backend be_tcp_{{$cfgIdx}}
  balance leastconn
  timeout check 5000ms
  stick-table type ip size 1m expire 5m{{ if (len $.PeerEndpoints) gt 0 }} peers openshift_peers {{ end }}
  stick on src
                {{ range $endpointID, $endpoint := $serviceUnit.EndpointTable }}
  server {{$endpointID}} {{$endpoint.IP}}:{{$endpoint.Port}} check inter 5000ms
                {{ end }}
            {{ end }}

After this modification, you can rebuild your router.

After you have made any desired modifications to the template, such as the example stick tables customization, you must rebuild your router for your changes to go in effect: