Adding a host to the cluster

For information on adding master or node hosts to a cluster, see the Adding hosts to an existing cluster section in the Install and configuration guide.

Master host tasks

Deprecating a master host

Deprecating a master host removes it from the OpenShift Origin environment.

The reasons to deprecate or scale down master hosts include hardware re-sizing or replacing the underlying infrastructure.

Highly available OpenShift Origin environments require at least three master hosts and three etcd nodes. Usually, the master hosts are collocated with the etcd services. This topic describes the deprecation process for master hosts with or without collocated etcd.

You should create a backup of the configuration and data files prior to any important task such as deprecating a master host. See the Creating a master host backup and etcd tasks sections for more information.

Ensure that the master and etcd services are always deployed in odd numbers due to the voting mechanisms that take place among those services.

Deprecating a master host without collocated etcd

Master hosts run important services, such as the OpenShift Origin API and controllers services (if multiple masters are present). In order to deprecate a master host, these services must be stopped.

The OpenShift Origin API service is an active/active service, so stopping the service does not affect the environment as long as the requests are sent to a separate master server. However, the OpenShift Origin controllers service is an active/passive service, where the services leverage etcd to decide the active master.

Deprecating a master host in a multi-master architecture includes removing the master from the load balancer pool to avoid new connections attempting to use that master. This process depends heavily on the load balancer used. The steps below show the details of removing the master from haproxy. In the event that OpenShift Origin is running on a cloud provider, or using a F5 appliance, see the specific product documents to remove the master from rotation.

Procedure

  1. Remove the backend section in the /etc/haproxy/haproxy.cfg configuration file. For example, if deprecating a master named master-0.example.com using haproxy, ensure the host name is removed from the following:

    backend mgmt8443
        balance source
        mode tcp
        # MASTERS 8443
        server master-1.example.com 192.168.55.12:8443 check
        server master-2.example.com 192.168.55.13:8443 check
  2. Then, restart the haproxy service.

    $ sudo systemctl restart haproxy
  3. Once the master is removed from the load balancer, disable the API and controller services:

    $ sudo systemctl disable --now atomic-openshift-master-api
    $ sudo systemctl disable --now atomic-openshift-master-controllers
  4. Because the master host is a unschedulable OpenShift Origin node, follow the steps in the Deprecating a node host section.

  5. Remove the master host from the [masters] and [nodes] groups in the /etc/ansible/hosts Ansible inventory file to avoid issues if running any Ansible tasks using that inventory file:

    ...[OUTPUT OMITTED]...
    # host group for masters
    [masters]
    *master-0.example.com*
    master-1.example.com
    master-2.example.com
    
    # host group for nodes, includes region info
    [nodes]
    *master-0.example.com openshift_node_labels="{'role': 'master'}" openshift_hostname=master-0.example.com openshift_schedulable=false*
    ...[OUTPUT OMITTED]...

    Deprecating the first master host listed in the Ansible inventory file requires extra precautions.

    At the time of writing, the /etc/origin/master/ca.serial.txt file is generated onto only the first master listed in the Ansible host inventory. This is being investigated to be fixed in future OpenShift Origin releases in the 1469358 bugzilla. If deprecating the first master host, copy the /etc/origin/master/ca.serial.txt file to the rest of master hosts before the process.

  6. The kubernetes service includes the master host IPs as endpoints. To verify that the master has been properly deprecated, review the kubernetes service output and see if the deprecated master has been removed:

    $ oc describe svc kubernetes -n default
    Name:			kubernetes
    Namespace:		default
    Labels:			component=apiserver
    			provider=kubernetes
    Annotations:		<none>
    Selector:		<none>
    Type:			ClusterIP
    IP:			10.111.0.1
    Port:			https	443/TCP
    Endpoints:		192.168.55.12:8443,192.168.55.13:8443
    Port:			dns	53/UDP
    Endpoints:		192.168.55.12:8053,192.168.55.13:8053
    Port:			dns-tcp	53/TCP
    Endpoints:		192.168.55.12:8053,192.168.55.13:8053
    Session Affinity:	ClientIP
    Events:			<none>

    Once the master has been successfully deprecated, the host where the master was previously running can be safely deleted.

Deprecating master host with collocated etcd

To deprecate a master host running an etcd service, execute the previous steps in Deprecating a master host without collocated etcd, as well as the steps in Removing an etcd host.

Replacing a master host

In the event of replacing a broken master host, follow the process in Deprecating a master host without collocated etcd, then scale up the master hosts using the scale up Ansible playbook following the steps in Adding hosts to an existing cluster.

If the master host has a collocated etcd, use the Deprecating master host with collocated etcd steps, then the Adding hosts to an existing cluster as well as Scaling etcd.

Creating a master host backup

The backup process is to be performed before any change to the infrastructure, such as a system update, upgrade, or any other significant modification. Backups should be performed on a regular basis to ensure the most recent data is available if a failure occurs.

OpenShift Origin files

The master instances run important services, such as the API, controllers. The /etc/origin/master directory stores many important files:

  • The configuration the API, controllers, services, and more

  • Certificates generated by the installation

  • All cloud provider-related configuration

  • Keys and other authentication files, such as htpasswd if using htpasswd

  • And more

The OpenShift Origin services can be customized to increase the log level, use proxies, and so on. The configuration files are stored in the /etc/sysconfig directory.

Because the masters are also unschedulable nodes, back up the entire /etc/origin directory.

Procedure

  1. Create a backup of the master host configuration files:

    $ MYBACKUPDIR=/backup/$(hostname)/$(date +%Y%m%d)
    $ sudo mkdir -p ${MYBACKUPDIR}/etc/sysconfig
    $ sudo cp -aR /etc/origin ${MYBACKUPDIR}/etc
    $ sudo cp -aR /etc/sysconfig/atomic-* ${MYBACKUPDIR}/etc/sysconfig/

    On a single master cluster installation the configuration file is stored in the /etc/sysconfig/atomic-openshift-master, whereas in a multi-master environment /etc/sysconfig/atomic-openshift-master-api, and /etc/sysconfig/atomic-openshift-master-controllers are used.

    At the time of writing, the /etc/origin/master/ca.serial.txt file is generated onto only the first master listed in the Ansible host inventory. This is being investigated to be fixed for future OpenShift Origin releases in the 1469358 bugzilla. If deprecating the first master host, copy the /etc/origin/master/ca.serial.txt file to the rest of master hosts before the process.

  2. Other important files that need to be considered when planning a backup include:

    File

    Description

    /etc/cni/*

    Container Network Interface configuration (if used)

    /etc/sysconfig/iptables

    Where the iptables rules are stored

    /etc/sysconfig/docker-storage-setup

    The input file for container-storage-setup command

    /etc/sysconfig/docker

    The docker configuration file

    /etc/sysconfig/docker-network

    docker networking configuration (i.e. MTU)

    /etc/sysconfig/docker-storage

    docker storage configuration (generated by container-storage-setup)

    /etc/dnsmasq.conf

    Main configuration file for dnsmasq

    /etc/dnsmasq.d/*

    Different dnsmasq configuration files

    /etc/sysconfig/flanneld

    flannel configuration file (if used)

    /etc/pki/ca-trust/source/anchors/

    Certificates added to the system (i.e. for external registries)

    Create a backup of those files:

    $ MYBACKUPDIR=/backup/$(hostname)/$(date +%Y%m%d)
    $ sudo mkdir -p ${MYBACKUPDIR}/etc/sysconfig
    $ sudo mkdir -p ${MYBACKUPDIR}/etc/pki/ca-trust/source/anchors
    $ sudo cp -aR /etc/sysconfig/{iptables,docker-*,flanneld} \
        ${MYBACKUPDIR}/etc/sysconfig/
    $ sudo cp -aR /etc/dnsmasq* /etc/cni ${MYBACKUPDIR}/etc/
    $ sudo cp -aR /etc/pki/ca-trust/source/anchors/* \
        ${MYBACKUPDIR}/etc/pki/ca-trust/source/anchors/
  3. If a package is accidentally removed, or a file included in an rpm package should be restored, having a list of rhel packages installed on the system can be useful.

    If using Red Hat Satellite features, such as content views or the facts store, provide a proper mechanism to reinstall the missing packages and a historical data of packages installed in the systems.

    To create a list of the current rhel packages installed in the system:

    $ MYBACKUPDIR=/backup/$(hostname)/$(date +%Y%m%d)
    $ sudo mkdir -p ${MYBACKUPDIR}
    $ rpm -qa | sort | sudo tee $MYBACKUPDIR/packages.txt
  4. If using the previous steps, the following files should now be present in the backup directory:

    $ MYBACKUPDIR=/backup/$(hostname)/$(date +%Y%m%d)
    $ sudo find ${MYBACKUPDIR} -mindepth 1 -type f -printf '%P\n'
    etc/sysconfig/atomic-openshift-master
    etc/sysconfig/atomic-openshift-master-api
    etc/sysconfig/atomic-openshift-master-controllers
    etc/sysconfig/atomic-openshift-node
    etc/sysconfig/flanneld
    etc/sysconfig/iptables
    etc/sysconfig/docker-network
    etc/sysconfig/docker-storage
    etc/sysconfig/docker-storage-setup
    etc/sysconfig/docker-storage-setup.rpmnew
    etc/origin/master/ca.crt
    etc/origin/master/ca.key
    etc/origin/master/ca.serial.txt
    etc/origin/master/ca-bundle.crt
    etc/origin/master/master.proxy-client.crt
    etc/origin/master/master.proxy-client.key
    etc/origin/master/service-signer.crt
    etc/origin/master/service-signer.key
    etc/origin/master/serviceaccounts.private.key
    etc/origin/master/serviceaccounts.public.key
    etc/origin/master/openshift-master.crt
    etc/origin/master/openshift-master.key
    etc/origin/master/openshift-master.kubeconfig
    etc/origin/master/master.server.crt
    etc/origin/master/master.server.key
    etc/origin/master/master.kubelet-client.crt
    etc/origin/master/master.kubelet-client.key
    etc/origin/master/admin.crt
    etc/origin/master/admin.key
    etc/origin/master/admin.kubeconfig
    etc/origin/master/etcd.server.crt
    etc/origin/master/etcd.server.key
    etc/origin/master/master.etcd-client.key
    etc/origin/master/master.etcd-client.csr
    etc/origin/master/master.etcd-client.crt
    etc/origin/master/master.etcd-ca.crt
    etc/origin/master/policy.json
    etc/origin/master/scheduler.json
    etc/origin/master/htpasswd
    etc/origin/master/session-secrets.yaml
    etc/origin/master/openshift-router.crt
    etc/origin/master/openshift-router.key
    etc/origin/master/registry.crt
    etc/origin/master/registry.key
    etc/origin/master/master-config.yaml
    etc/origin/generated-configs/master-master-1.example.com/master.server.crt
    ...[OUTPUT OMITTED]...
    etc/origin/cloudprovider/openstack.conf
    etc/origin/node/system:node:master-0.example.com.crt
    etc/origin/node/system:node:master-0.example.com.key
    etc/origin/node/ca.crt
    etc/origin/node/system:node:master-0.example.com.kubeconfig
    etc/origin/node/server.crt
    etc/origin/node/server.key
    etc/origin/node/node-dnsmasq.conf
    etc/origin/node/resolv.conf
    etc/origin/node/node-config.yaml
    etc/origin/node/flannel.etcd-client.key
    etc/origin/node/flannel.etcd-client.csr
    etc/origin/node/flannel.etcd-client.crt
    etc/origin/node/flannel.etcd-ca.crt
    etc/pki/ca-trust/source/anchors/openshift-ca.crt
    etc/pki/ca-trust/source/anchors/registry-ca.crt
    etc/dnsmasq.conf
    etc/dnsmasq.d/origin-dns.conf
    etc/dnsmasq.d/origin-upstream-dns.conf
    etc/dnsmasq.d/node-dnsmasq.conf
    packages.txt

    If needed, the files can be compressed to save space:

    $ MYBACKUPDIR=*/backup/$(hostname)/$(date +%Y%m%d)*
    $ sudo tar -zcvf */backup/*$(hostname)-$(date +%Y%m%d).tar.gz $MYBACKUPDIR
    $ sudo rm -Rf ${MYBACKUPDIR}

To create any of these files from scratch, the openshift-ansible-contrib repository contains the backup_master_node.sh script, which performs the previous steps. The script creates a directory on the host running the script and copies all the files previously mentioned.

The openshift-ansible-contrib script is not supported by Red Hat, but the reference architecture team performs testing to ensure the code operates as defined and is secure.

The script can be executed on every master host with:

$ mkdir ~/git
$ cd ~/git
$ git clone https://github.com/openshift/openshift-ansible-contrib.git
$ cd openshift-ansible-contrib/reference-architecture/day2ops/scripts
$ ./backup_master_node.sh -h

Restoring a master host backup

After creating a backup of important master host files, if they become corrupted or accidentally removed, you can restore the file by copying back the file, ensuring it contains the proper content and restart the affected services.

Procedure

  1. Restore the /etc/origin/master/master-config.yaml file:

    # MYBACKUPDIR=*/backup/$(hostname)/$(date +%Y%m%d)*
    # cp /etc/origin/master/master-config.yaml /etc/origin/master/master-config.yaml.old
    # cp /backup/$(hostname)/$(date +%Y%m%d)/origin/master/master-config.yaml /etc/origin/master/master-config.yaml
    # systemctl restart atomic-openshift-master-api
    # systemctl restart atomic-openshift-master-controllers

    Restarting the master services can lead to downtime. However, you can remove the master host from the highly available load balancer pool, then perform the restore operation. Once the service has been properly restored, you can add the master host back to the load balancer pool.

    Perform a full reboot of the affected instance to restore the iptables configuration.

  2. If the issue is an accidental package and its dependencies are removed, reinstall the package.

    Get the list of the current installed packages:

    $ rpm -qa | sort > /tmp/current_packages.txt
  3. Get the differences:

    $ diff /tmp/current_packages.txt ${MYBACKUPDIR}/packages.txt
    1a2
    > ansible-2.4.0.0-5.el7.noarch
  4. Reinstall the missing packages:

    # yum reinstall -y *ansible-2.4.0.0-5.el7.noarch*
  5. Restore a system certificate by copying the certificate to the /etc/pki/ca-trust/source/anchors/ directory and execute the update-ca-trust:

    $ MYBACKUPDIR=*/backup/$(hostname)/$(date +%Y%m%d)*
    $ sudo cp ${MYBACKUPDIR}/external_certificates/my_company.crt /etc/pki/ca-trust/source/anchors/
    $ sudo update-ca-trust

    Always ensure the user ID and group ID are restored when the files are copied back, as well as the SELinux context.

Node host tasks

Deprecating a node host

The procedure is the same whether deprecating an infrastructure node or an application node.

Prerequisites

Ensure enough capacity is available to migrate the existing pods from the node set to be removed. Removing an infrastructure node is advised only when at least two more nodes will stay online after the infrastructure node is removed.

Procedure

  1. List all available nodes to find the node to deprecate:

    $ oc get nodes
    NAME                  STATUS                     AGE       VERSION
    ocp-infra-node-b7pl   Ready                      23h       v1.6.1+5115d708d7
    ocp-infra-node-p5zj   Ready                      23h       v1.6.1+5115d708d7
    ocp-infra-node-rghb   Ready                      23h       v1.6.1+5115d708d7
    ocp-master-dgf8       Ready,SchedulingDisabled   23h       v1.6.1+5115d708d7
    ocp-master-q1v2       Ready,SchedulingDisabled   23h       v1.6.1+5115d708d7
    ocp-master-vq70       Ready,SchedulingDisabled   23h       v1.6.1+5115d708d7
    ocp-node-020m         Ready                      23h       v1.6.1+5115d708d7
    ocp-node-7t5p         Ready                      23h       v1.6.1+5115d708d7
    ocp-node-n0dd         Ready                      23h       v1.6.1+5115d708d7

    As an example, this topic deprecates the ocp-infra-node-b7pl infrastructure node.

  2. Describe the node and its running services:

    $ oc describe node ocp-infra-node-b7pl
    Name:			ocp-infra-node-b7pl
    Role:
    Labels:			beta.kubernetes.io/arch=amd64
    			beta.kubernetes.io/instance-type=n1-standard-2
    			beta.kubernetes.io/os=linux
    			failure-domain.beta.kubernetes.io/region=europe-west3
    			failure-domain.beta.kubernetes.io/zone=europe-west3-c
    			kubernetes.io/hostname=ocp-infra-node-b7pl
    			role=infra
    Annotations:		volumes.kubernetes.io/controller-managed-attach-detach=true
    Taints:			<none>
    CreationTimestamp:	Wed, 22 Nov 2017 09:36:36 -0500
    Phase:
    Conditions:
      ...
    Addresses:		10.156.0.11,ocp-infra-node-b7pl
    Capacity:
     cpu:		2
     memory:	7494480Ki
     pods:		20
    Allocatable:
     cpu:		2
     memory:	7392080Ki
     pods:		20
    System Info:
     Machine ID:			bc95ccf67d047f2ae42c67862c202e44
     System UUID:			9762CC3D-E23C-AB13-B8C5-FA16F0BCCE4C
     Boot ID:			ca8bf088-905d-4ec0-beec-8f89f4527ce4
     Kernel Version:		3.10.0-693.5.2.el7.x86_64
     OS Image:			Employee SKU
     Operating System:		linux
     Architecture:			amd64
     Container Runtime Version:	docker://1.12.6
     Kubelet Version:		v1.6.1+5115d708d7
     Kube-Proxy Version:		v1.6.1+5115d708d7
    ExternalID:			437740049672994824
    Non-terminated Pods:		(2 in total)
      Namespace			Name				CPU Requests	CPU Limits	Memory Requests	Memory Limits
      ---------			----				------------	----------	---------------	-------------
      default			docker-registry-1-5szjs		100m (5%)	0 (0%)		256Mi (3%)0 (0%)
      default			router-1-vzlzq			100m (5%)	0 (0%)		256Mi (3%)0 (0%)
    Allocated resources:
      (Total limits may be over 100 percent, i.e., overcommitted.)
      CPU Requests	CPU Limits	Memory Requests	Memory Limits
      ------------	----------	---------------	-------------
      200m (10%)	0 (0%)		512Mi (7%)	0 (0%)
    Events:		<none>

    The output above shows that the node is running two pods: router-1-vzlzq and docker-registry-1-5szjs. Two more infrastructure nodes are available to migrate these two pods.

    The cluster described above is a highly available cluster, this means both the router and docker-registry services are running on all infrastructure nodes.

  3. Mark a node as unschedulable and evacuate all of its pods:

    $ oc adm drain ocp-infra-node-b7pl --delete-local-data
    node "ocp-infra-node-b7pl" cordoned
    WARNING: Deleting pods with local storage: docker-registry-1-5szjs
    pod "docker-registry-1-5szjs" evicted
    pod "router-1-vzlzq" evicted
    node "ocp-infra-node-b7pl" drained

    If the pod has attached local storage (for example, EmptyDir), the --delete-local-data option must be provided. Generally, pods running in production should use the local storage only for temporary or cache files, but not for anything important or persistent. For regular storage, applications should use object storage or persistent volumes. In this case, the docker-registry pod’s local storage is empty, because the object storage is used instead to store the container images.

    The above operation deletes existing pods running on the node. Then, new pods are created according to the replication controller.

    In general, every application should be deployed with a deployment configuration, which creates pods using the replication controller.

    oc adm drain will not delete any bare pods (pods that are neither mirror pods nor managed by ReplicationController, ReplicaSet, DaemonSet, StatefulSet, or a job). To do so, the --force option is required. Be aware that the bare pods will not be recreated on other nodes and data may be lost during this operation.

    The example below shows the output of the replication controller of the registry:

    $ oc describe rc/docker-registry-1
    Name:		docker-registry-1
    Namespace:	default
    Selector:	deployment=docker-registry-1,deploymentconfig=docker-registry,docker-registry=default
    Labels:		docker-registry=default
    		openshift.io/deployment-config.name=docker-registry
    Annotations: ...
    Replicas:	3 current / 3 desired
    Pods Status:	3 Running / 0 Waiting / 0 Succeeded / 0 Failed
    Pod Template:
      Labels:		deployment=docker-registry-1
    			deploymentconfig=docker-registry
    			docker-registry=default
      Annotations:		openshift.io/deployment-config.latest-version=1
    			openshift.io/deployment-config.name=docker-registry
    			openshift.io/deployment.name=docker-registry-1
      Service Account:	registry
      Containers:
       registry:
        Image:	openshift3/ose-docker-registry:v3.6.173.0.49
        Port:	5000/TCP
        Requests:
          cpu:	100m
          memory:	256Mi
        Liveness:	http-get https://:5000/healthz delay=10s timeout=5s period=10s #success=1 #failure=3
        Readiness:	http-get https://:5000/healthz delay=0s timeout=5s period=10s #success=1 #failure=3
        Environment:
          REGISTRY_HTTP_ADDR:					:5000
          REGISTRY_HTTP_NET:					tcp
          REGISTRY_HTTP_SECRET:					tyGEnDZmc8dQfioP3WkNd5z+Xbdfy/JVXf/NLo3s/zE=
          REGISTRY_MIDDLEWARE_REPOSITORY_OPENSHIFT_ENFORCEQUOTA:	false
          REGISTRY_HTTP_TLS_KEY:					/etc/secrets/registry.key
          OPENSHIFT_DEFAULT_REGISTRY:				docker-registry.default.svc:5000
          REGISTRY_CONFIGURATION_PATH:				/etc/registry/config.yml
          REGISTRY_HTTP_TLS_CERTIFICATE:				/etc/secrets/registry.crt
        Mounts:
          /etc/registry from docker-config (rw)
          /etc/secrets from registry-certificates (rw)
          /registry from registry-storage (rw)
      Volumes:
       registry-storage:
        Type:	EmptyDir (a temporary directory that shares a pod's lifetime)
        Medium:
       registry-certificates:
        Type:	Secret (a volume populated by a Secret)
        SecretName:	registry-certificates
        Optional:	false
       docker-config:
        Type:	Secret (a volume populated by a Secret)
        SecretName:	registry-config
        Optional:	false
    Events:
      FirstSeen	LastSeen	Count	From			SubObjectPath	Type		Reason		Message
      ---------	--------	-----	----			-------------	--------	------		-------
      49m		49m		1	replication-controller			Normal		SuccessfulCreate	Created pod: docker-registry-1-dprp5

    The event at the bottom of the output displays information about new pod creation. So, when listing all pods:

    $ oc get pods
    NAME                       READY     STATUS    RESTARTS   AGE
    docker-registry-1-dprp5    1/1       Running   0          52m
    docker-registry-1-kr8jq    1/1       Running   0          1d
    docker-registry-1-ncpl2    1/1       Running   0          1d
    registry-console-1-g4nqg   1/1       Running   0          1d
    router-1-2gshr             0/1       Pending   0          52m
    router-1-85qm4             1/1       Running   0          1d
    router-1-q5sr8             1/1       Running   0          1d
  4. The docker-registry-1-5szjs and router-1-vzlzq pods that were running on the now deprecated node are no longer available. Instead, two new pods have been created: docker-registry-1-dprp5 and router-1-2gshr. As shown above, the new router pod is router-1-2gshr, but is in the Pending state. This is because every node can be running only on one single router and is bound to the ports 80 and 443 of the host.

  5. When observing the newly created registry pod, the example below shows that the pod has been created on the ocp-infra-node-rghb node, which is different from the deprecating node:

    $ oc describe pod docker-registry-1-dprp5
    Name:			docker-registry-1-dprp5
    Namespace:		default
    Security Policy:	hostnetwork
    Node:			ocp-infra-node-rghb/10.156.0.10
    ...

    The only difference between deprecating the infrastructure and the application node is that once the infrastructure node is evacuated, and if there is no plan to replace that node, the services running on infrastructure nodes can be scaled down:

    $ oc scale dc/router --replicas 2
    deploymentconfig "router" scaled
    
    $ oc scale dc/docker-registry --replicas 2
    deploymentconfig "docker-registry" scaled
  6. Now, every infrastructure node is running only one kind of each pod:

    $ oc get pods
    NAME                       READY     STATUS    RESTARTS   AGE
    docker-registry-1-kr8jq    1/1       Running   0          1d
    docker-registry-1-ncpl2    1/1       Running   0          1d
    registry-console-1-g4nqg   1/1       Running   0          1d
    router-1-85qm4             1/1       Running   0          1d
    router-1-q5sr8             1/1       Running   0          1d
    
    $ oc describe po/docker-registry-1-kr8jq | grep Node:
    Node:			ocp-infra-node-p5zj/10.156.0.9
    
    $ oc describe po/docker-registry-1-ncpl2 | grep Node:
    Node:			ocp-infra-node-rghb/10.156.0.10

    To provide a full highly available cluster, at least three infrastructure nodes should always be available.

  7. To verify that the scheduling on the node is disabled:

    $ oc get nodes
    NAME                  STATUS                     AGE       VERSION
    ocp-infra-node-b7pl   Ready,SchedulingDisabled   1d        v1.6.1+5115d708d7
    ocp-infra-node-p5zj   Ready                      1d        v1.6.1+5115d708d7
    ocp-infra-node-rghb   Ready                      1d        v1.6.1+5115d708d7
    ocp-master-dgf8       Ready,SchedulingDisabled   1d        v1.6.1+5115d708d7
    ocp-master-q1v2       Ready,SchedulingDisabled   1d        v1.6.1+5115d708d7
    ocp-master-vq70       Ready,SchedulingDisabled   1d        v1.6.1+5115d708d7
    ocp-node-020m         Ready                      1d        v1.6.1+5115d708d7
    ocp-node-7t5p         Ready                      1d        v1.6.1+5115d708d7
    ocp-node-n0dd         Ready                      1d        v1.6.1+5115d708d7

    And that the node does not contain any pods:

    $ oc describe node ocp-infra-node-b7pl
    Name:			ocp-infra-node-b7pl
    Role:
    Labels:			beta.kubernetes.io/arch=amd64
    			beta.kubernetes.io/instance-type=n1-standard-2
    			beta.kubernetes.io/os=linux
    			failure-domain.beta.kubernetes.io/region=europe-west3
    			failure-domain.beta.kubernetes.io/zone=europe-west3-c
    			kubernetes.io/hostname=ocp-infra-node-b7pl
    			role=infra
    Annotations:		volumes.kubernetes.io/controller-managed-attach-detach=true
    Taints:			<none>
    CreationTimestamp:	Wed, 22 Nov 2017 09:36:36 -0500
    Phase:
    Conditions:
      ...
    Addresses:		10.156.0.11,ocp-infra-node-b7pl
    Capacity:
     cpu:		2
     memory:	7494480Ki
     pods:		20
    Allocatable:
     cpu:		2
     memory:	7392080Ki
     pods:		20
    System Info:
     Machine ID:			bc95ccf67d047f2ae42c67862c202e44
     System UUID:			9762CC3D-E23C-AB13-B8C5-FA16F0BCCE4C
     Boot ID:			ca8bf088-905d-4ec0-beec-8f89f4527ce4
     Kernel Version:		3.10.0-693.5.2.el7.x86_64
     OS Image:			Employee SKU
     Operating System:		linux
     Architecture:			amd64
     Container Runtime Version:	docker://1.12.6
     Kubelet Version:		v1.6.1+5115d708d7
     Kube-Proxy Version:		v1.6.1+5115d708d7
    ExternalID:			437740049672994824
    Non-terminated Pods:		(0 in total)
      Namespace			Name		CPU Requests	CPU Limits	Memory Requests	Memory Limits
      ---------			----		------------	----------	---------------	-------------
    Allocated resources:
      (Total limits may be over 100 percent, i.e., overcommitted.)
      CPU Requests	CPU Limits	Memory Requests	Memory Limits
      ------------	----------	---------------	-------------
      0 (0%)	0 (0%)		0 (0%)		0 (0%)
    Events:		<none>
  8. Remove the infrastructure instance from the backend section in the /etc/haproxy/haproxy.cfg configuration file:

    backend router80
        balance source
        mode tcp
        server infra-1.example.com 192.168.55.12:80 check
        server infra-2.example.com 192.168.55.13:80 check
    
    backend router443
        balance source
        mode tcp
        server infra-1.example.com 192.168.55.12:443 check
        server infra-2.example.com 192.168.55.13:443 check
  9. Then, restart the haproxy service.

    $ sudo systemctl restart haproxy
  10. Remove the node from the cluster after all pods are evicted with command:

    $ oc delete node ocp-infra-node-b7pl
    node "ocp-infra-node-b7pl" deleted
    $ oc get nodes
    NAME                  STATUS                     AGE       VERSION
    ocp-infra-node-p5zj   Ready                      1d        v1.6.1+5115d708d7
    ocp-infra-node-rghb   Ready                      1d        v1.6.1+5115d708d7
    ocp-master-dgf8       Ready,SchedulingDisabled   1d        v1.6.1+5115d708d7
    ocp-master-q1v2       Ready,SchedulingDisabled   1d        v1.6.1+5115d708d7
    ocp-master-vq70       Ready,SchedulingDisabled   1d        v1.6.1+5115d708d7
    ocp-node-020m         Ready                      1d        v1.6.1+5115d708d7
    ocp-node-7t5p         Ready                      1d        v1.6.1+5115d708d7
    ocp-node-n0dd         Ready                      1d        v1.6.1+5115d708d7

For more information on evacuating and draining pods or nodes, see Node maintenance section.

Replacing a node host

In the event that a node would need to be added in place of the deprecated node, follow the Adding hosts to an existing cluster section.

Creating a node host backup

The backup process is to be performed before any change to the infrastructure, such as a system update, upgrade, or any other significant modification. Backups should be performed on a regular basis to ensure the most recent data is available if a failure occurs.

OpenShift Origin files

Node instances run applications in the form of pods, which are based on containers. The /etc/origin/ and /etc/origin/node directories house important files, such as:

  • The configuration of the node services

  • Certificates generated by the installation

  • Cloud provider-related configuration

  • Keys and other authentication files, such as the dnsmasq configuration

The OpenShift Origin services can be customized to increase the log level, use proxies, and more, and the configuration files are stored in the /etc/sysconfig directory.

Procedure

  1. Create a backup of the node configuration files:

    $ MYBACKUPDIR=/backup/$(hostname)/$(date +%Y%m%d)
    $ sudo mkdir -p ${MYBACKUPDIR}/etc/sysconfig
    $ sudo cp -aR /etc/origin ${MYBACKUPDIR}/etc
    $ sudo cp -aR /etc/sysconfig/atomic-openshift-node ${MYBACKUPDIR}/etc/sysconfig/
  2. OpenShift Origin uses specific files that must be taken into account when planning the backup policy, including:

    File

    Description

    /etc/cni/*

    Container Network Interface configuration (if used)

    /etc/sysconfig/iptables

    Where the iptables rules are stored

    /etc/sysconfig/docker-storage-setup

    The input file for container-storage-setup command

    /etc/sysconfig/docker

    The docker configuration file

    /etc/sysconfig/docker-network

    docker networking configuration (i.e. MTU)

    /etc/sysconfig/docker-storage

    docker storage configuration (generated by container-storage-setup)

    /etc/dnsmasq.conf

    Main configuration file for dnsmasq

    /etc/dnsmasq.d/*

    Different dnsmasq configuration files

    /etc/sysconfig/flanneld

    flannel configuration file (if used)

    /etc/pki/ca-trust/source/anchors/

    Certificates added to the system (i.e. for external registries)

    To create those files:

    $ MYBACKUPDIR=/backup/$(hostname)/$(date +%Y%m%d)
    $ sudo mkdir -p ${MYBACKUPDIR}/etc/sysconfig
    $ sudo mkdir -p ${MYBACKUPDIR}/etc/pki/ca-trust/source/anchors
    $ sudo cp -aR /etc/sysconfig/{iptables,docker-*,flanneld} \
        ${MYBACKUPDIR}/etc/sysconfig/
    $ sudo cp -aR /etc/dnsmasq* /etc/cni ${MYBACKUPDIR}/etc/
    $ sudo cp -aR /etc/pki/ca-trust/source/anchors/* \
        ${MYBACKUPDIR}/etc/pki/ca-trust/source/anchors/
  3. If a package is accidentally removed, or a file included in an rpm package should be restored, having a list of rhel packages installed on the system can be useful.

    If using Red Hat Satellite features, such as content views or the facts store, provide a proper mechanism to reinstall the missing packages and a historical data of packages installed in the systems.

    To create a list of the current rhel packages installed in the system:

    $ MYBACKUPDIR=/backup/$(hostname)/$(date +%Y%m%d)
    $ sudo mkdir -p ${MYBACKUPDIR}
    $ rpm -qa | sort | sudo tee $MYBACKUPDIR/packages.txt
  4. The following files should now be present in the backup directory:

    $ MYBACKUPDIR=*/backup/$(hostname)/$(date +%Y%m%d)*
    $ sudo find ${MYBACKUPDIR} -mindepth 1 -type f -printf '%P\n'
    etc/sysconfig/atomic-openshift-node
    etc/sysconfig/flanneld
    etc/sysconfig/iptables
    etc/sysconfig/docker-network
    etc/sysconfig/docker-storage
    etc/sysconfig/docker-storage-setup
    etc/sysconfig/docker-storage-setup.rpmnew
    etc/origin/node/system:node:app-node-0.example.com.crt
    etc/origin/node/system:node:app-node-0.example.com.key
    etc/origin/node/ca.crt
    etc/origin/node/system:node:app-node-0.example.com.kubeconfig
    etc/origin/node/server.crt
    etc/origin/node/server.key
    etc/origin/node/node-dnsmasq.conf
    etc/origin/node/resolv.conf
    etc/origin/node/node-config.yaml
    etc/origin/node/flannel.etcd-client.key
    etc/origin/node/flannel.etcd-client.csr
    etc/origin/node/flannel.etcd-client.crt
    etc/origin/node/flannel.etcd-ca.crt
    etc/origin/cloudprovider/openstack.conf
    etc/pki/ca-trust/source/anchors/openshift-ca.crt
    etc/pki/ca-trust/source/anchors/registry-ca.crt
    etc/dnsmasq.conf
    etc/dnsmasq.d/origin-dns.conf
    etc/dnsmasq.d/origin-upstream-dns.conf
    etc/dnsmasq.d/node-dnsmasq.conf
    packages.txt

    If needed, the files can be compressed to save space:

    $ MYBACKUPDIR=*/backup/$(hostname)/$(date +%Y%m%d)*
    $ sudo tar -zcvf */backup/*$(hostname)-$(date +%Y%m%d).tar.gz $MYBACKUPDIR
    $ sudo rm -Rf ${MYBACKUPDIR}

To create any of these files from scratch, the openshift-ansible-contrib repository contains the backup_master_node.sh script, which performs the previous steps. The script creates a directory on the host running the script and copies all the files previously mentioned.

The openshift-ansible-contrib script is not supported by Red Hat, but the reference architecture team performs testing to ensure the code operates as defined and is secure.

The script can be executed on every master host with:

$ mkdir ~/git
$ cd ~/git
$ git clone https://github.com/openshift/openshift-ansible-contrib.git
$ cd openshift-ansible-contrib/reference-architecture/day2ops/scripts
$ ./backup_master_node.sh -h

Restoring a node host backup

After creating a backup of important node host files, if they become corrupted or accidentally removed, you can restore the file by copying back the file, ensuring it contains the proper content and restart the affected services.

Procedure

  1. Restore the /etc/origin/node/node-config.yaml file:

    # MYBACKUPDIR=/backup/$(hostname)/$(date +%Y%m%d)
    # cp /etc/origin/node/node-config.yaml /etc/origin/node/node-config.yaml.old
    # cp /backup/$(hostname)/$(date +%Y%m%d)/etc/origin/node/node-config.yaml /etc/origin/node/node-config.yaml
    # systemctl restart atomic-openshift-node

Restarting the services can lead to downtime. See Node maintenance, for tips on how to ease the process.

Perform a full reboot of the affected instance to restore the iptables configuration.

  1. If the issue is an accidental package and its dependencies are removed, reinstall the package.

    Get the list of the current installed packages:

    $ rpm -qa | sort > /tmp/current_packages.txt
  2. Get the differences:

    $ diff /tmp/current_packages.txt ${MYBACKUPDIR}/packages.txt
    1a2
    > ansible-2.4.0.0-5.el7.noarch
  3. Reinstall the missing packages:

    # yum reinstall -y ansible-2.4.0.0-5.el7.noarch
  4. Restore a system certificate by copying the certificate to the /etc/pki/ca-trust/source/anchors/ directory and execute the update-ca-trust:

    $ MYBACKUPDIR=*/backup/$(hostname)/$(date +%Y%m%d)*
    $ sudo cp ${MYBACKUPDIR}/etc/pki/ca-trust/source/anchors/my_company.crt /etc/pki/ca-trust/source/anchors/
    $ sudo update-ca-trust

    Always ensure proper user ID and group ID are restored when the files are copied back, as well as the SELinux context.

Node maintenance

See Managing nodes or Managing pods topics for various node management options. These include:

etcd tasks

etcd is the key value store for all object definitions, as well as the persistent master state. Other components watch for changes, then bring themselves into the desired state.

OpenShift Origin versions prior to 3.5 used etcd version 2 (v2), while 3.5 and later use version 3 (v3). The data model between the two versions of etcd is different. etcd v3 can use both the v2 and v3 data model, whereas etcd v2 can only use the v2 data model. In an etcd v3 server, the v2 and v3 data stores exist in parallel and are independent.

For both v2 and v3 operations, you can use the ETCDCTL_API environment variable to use the proper API:

$ etcdctl -v
etcdctl version: 3.2.5
API version: 2
$ ETCDCTL_API=3 etcdctl version
etcdctl version: 3.2.5
API version: 3.2

See Migrating etcd Data (v2 to v3) section in the OpenShift Origin 3.7 documentation for information about how to migrate to v3.

Back up and restore etcd

The etcd backup process is composed of two different procedures:

  • Configuration backup: Including the required etcd configuration and certificates

  • Data backup: Including both v2 and v3 data model.

The data backup procedure can be done on any host that has connectivity to the etcd cluster, where the proper certificates are provided, and where the etcdctl tool is installed.

The backup files must be copied to an external system, ideally outside the OpenShift Origin environment, and then encrypted.

etcd configuration backup

The etcd configuration files to be preserved are all stored in the /etc/etcd directory of the instances where etcd is running. This includes the etcd configuration file (/etc/etcd/etcd.conf) and the required certificates for cluster communication. All those files are generated at installation time by the Ansible installer.

To back up the etcd configuration:

$ ssh master-0
# mkdir -p /backup/etcd-config-$(date +%Y%m%d)/
# cp -R /etc/etcd/ /backup/etcd-config-$(date +%Y%m%d)/

The backup is to be performed on every etcd member of the cluster as the certificates and configuration files are unique.

etcd configuration restore

The restore procedure for etcd configuration files replaces the appropriate files, then restarts the service.

If an etcd host has become corrupted and the /etc/etcd/etcd.conf file is lost, restore it using:

$ ssh master-0
# cp /backup/yesterday/master-0-files/etcd.conf /etc/etcd/etcd.conf
# restorecon -Rv /etc/etcd/etcd.conf
# systemctl restart etcd.service

In this example, the backup file is stored in the /backup/yesterday/master-0-files/etcd.conf path where it can be used as an external NFS share, S3 bucket, etc.

etcd data backup

Prerequisites

The OpenShift Origin installer creates aliases to avoid typing all the flags named etcdctl2 for etcd v2 tasks and etcdctl3 for etcd v3 tasks.

However, the etcdctl3 alias does not provide the full endpoint list to the etcdctl command, so the --endpoints option with all the endpoints must be provided.

Before backing up etcd:

  • etcdctl binaries should be available or, in containerized installations, the rhel7/etcd container should be available

  • Ensure connectivity with the etcd cluster (port 2379/tcp)

  • Ensure the proper certificates to connect to the etcd cluster

    1. To ensure the etcd cluster is working, check its health:

      # etcdctl --cert-file=/etc/etcd/peer.crt \
                --key-file=/etc/etcd/peer.key \
                --ca-file=/etc/etcd/ca.crt \
                --peers="https://*master-0.example.com*:2379,\
                https://*master-1.example.com*:2379,\
                https://*master-2.example.com*:2379"\
                cluster-health
      member 5ee217d19001 is healthy: got healthy result from https://192.168.55.12:2379
      member 2a529ba1840722c0 is healthy: got healthy result from https://192.168.55.8:2379
      member ed4f0efd277d7599 is healthy: got healthy result from https://192.168.55.13:2379
      cluster is healthy

      Or if using the etcd v3 API:

      # ETCDCTL_API=3 etcdctl --cert="/etc/etcd/peer.crt" \
                --key=/etc/etcd/peer.key \
                --cacert="/etc/etcd/ca.crt" \
                --endpoints="https://*master-0.example.com*:2379,\
                  https://*master-1.example.com*:2379,\
                  https://*master-2.example.com*:2379"
                  endpoint health
      https://master-0.example.com:2379 is healthy: successfully committed proposal: took = 5.011358ms
      https://master-1.example.com:2379 is healthy: successfully committed proposal: took = 1.305173ms
      https://master-2.example.com:2379 is healthy: successfully committed proposal: took = 1.388772ms
    2. Check the member list:

      # etcdctl2 member list
      2a371dd20f21ca8d: name=master-1.example.com peerURLs=https://192.168.55.12:2380 clientURLs=https://192.168.55.12:2379 isLeader=false
      40bef1f6c79b3163: name=master-0.example.com peerURLs=https://192.168.55.8:2380 clientURLs=https://192.168.55.8:2379 isLeader=false
      95dc17ffcce8ee29: name=master-2.example.com peerURLs=https://192.168.55.13:2380 clientURLs=https://192.168.55.13:2379 isLeader=true

      Or, if using etcd the v3 API:

      # etcdctl3 member list
      2a371dd20f21ca8d, started, master-1.example.com, https://192.168.55.12:2380, https://192.168.55.12:2379
      40bef1f6c79b3163, started, master-0.example.com, https://192.168.55.8:2380, https://192.168.55.8:2379
      95dc17ffcce8ee29, started, master-2.example.com, https://192.168.55.13:2380, https://192.168.55.13:2379

Procedure

While the etcdctl backup command is used to perform the backup, etcd v3 has no concept of a "backup". Instead, a "snapshot" may either be taken from a live member with the etcdctl snapshot save command or by copying the member/snap/db file from an etcd data directory.

The etcdctl backup command rewrites some of the metadata contained in the backup (specifically, the node ID and cluster ID), which means that in the backup, the node loses its former identity. In order to recreate a cluster from the backup, a new, single-node cluster is created, then the rest of the nodes join the cluster. The metadata is rewritten to prevent the new node from inadvertently being joined onto an existing cluster.

  1. Perform the backup:

    # mkdir -p */backup/etcd-$(date +%Y%m%d)*
    # systemctl stop etcd.service
    # etcdctl2 backup \
        --data-dir /var/lib/etcd \
        --backup-dir */backup/etcd-$(date +%Y%m%d)*
    # cp /var/lib/etcd/member/snap/db */backup/etcd-$(date +%Y%m%d)*
    # systemctl start etcd.service

    While stopping the etcd service is not strictly necessary, doing so ensures that the etcd data is fully synchronized.

    The etcdctl2 backup command creates etcd v2 data backup where copying the db file while the etcd service is not running is equivalent to running etcdctl3 snapshot for etcd v3 data backup:

    # mkdir -p */backup/etcd-$(date +%Y%m%d)*
    # etcdctl3 snapshot save */backup/etcd-$(date +%Y%m%d)*/db
    Snapshot saved at /backup/etcd-<date>/db
    # systemctl stop etcd.service
    # etcdctl2 backup \
        --data-dir /var/lib/etcd \
        --backup-dir */backup/etcd-$(date +%Y%m%d)*
    # systemctl start etcd.service

    The etcdctl snapshot save command requires the etcd service to be running.

    In this example, a /backup/etcd-<date>/ directory is created, where <date> represents the current date, which must be an external NFS share, S3 bucket, or any external storage location.

    In the case of an all-in-one cluster, the etcd data directory is located in /var/lib/origin/openshift.local.etcd

Restoring etcd v2 & v3 data

The following restores healthy data files and starts the etcd cluster as a single node, then adds the rest of the nodes in case an etcd cluster is required.

Procedure

  1. Stop all etcd services:

    # systemctl stop etcd.service
  2. Clean the etcd data directories to ensure the proper backup is restored, but keeping the running copy:

    # mv /var/lib/etcd /var/lib/etcd.old
    # mkdir /var/lib/etcd
    # chown -R etcd.etcd /var/lib/etcd/
    # restorecon -Rv /var/lib/etcd/

    Alternatively, you can wipe the etcd data directory:

    # rm -Rf /var/lib/etcd/*

    In case an all-in-one cluster, the etcd data directory is located in /var/lib/origin/openshift.local.etcd

  3. Restore a healthy backup data file to one of the etcd nodes:

    # cp -R /backup/etcd-xxx/* /var/lib/etcd/
    # mv /var/lib/etcd/db /var/lib/etcd/member/snap/db

    Perform this step on all etcd hosts (including master hosts collocated with etcd).

  4. Run the etcd service, forcing a new cluster.

    This creates a custom file for the etcd service, which overwrites the execution command adding the --force-new-cluster option:

    # mkdir -p /etc/systemd/system/etcd.service.d/
    # echo "[Service]" > /etc/systemd/system/etcd.service.d/temp.conf
    # echo "ExecStart=" >> /etc/systemd/system/etcd.service.d/temp.conf
    # sed -n '/ExecStart/s/"$/ --force-new-cluster"/p' \
        /usr/lib/systemd/system/etcd.service \
        >> /etc/systemd/system/etcd.service.d/temp.conf
    
    # systemctl daemon-reload
    # systemctl restart etcd
  5. Check for error messages:

    $ journalctl -fu etcd.service
  6. Check for health status (in this case, a single node):

    # etcdctl2 cluster-health
    member 5ee217d17301 is healthy: got healthy result from https://192.168.55.8:2379
    cluster is healthy
  7. Restart the etcd service in cluster mode:

    # rm -f /etc/systemd/system/etcd.service.d/temp.conf
    # systemctl daemon-reload
    # systemctl restart etcd
  8. Check for health status and member list

    # etcdctl2 cluster-health
    member 5ee217d17301 is healthy: got healthy result from https://192.168.55.8:2379
    cluster is healthy
    
    # etcdctl2 member list
    5ee217d17301: name=master-0.example.com peerURLs=http://localhost:2380 clientURLs=https://192.168.55.8:2379 isLeader=true
  9. Once the first instance is running, it is safe to restore multiple etcd servers as desired.

Fix the peerURLS parameter

After restoring the data and creating a new cluster, the peerURLs parameter shows localhost instead the IP where etcd is listening for peer communication:

# etcdctl2 member list
5ee217d17301: name=master-0.example.com peerURLs=http://*localhost*:2380 clientURLs=https://192.168.55.8:2379 isLeader=true

Procedure

  1. Get the member ID from the etcdctl member list output.

  2. Get the IP where etcd is listening for peer communication:

    $ ss -l4n | grep 2380
  3. Update the member information with that IP:

    # etcdctl2 member update *5ee217d17301* https://*192.168.55.8*:2380
    Updated member with ID 5ee217d17301 in cluster
  4. To verify, check that the IP is in the output of the following:

    $ etcdctl2 member list
    5ee217d17301: name=master-0.example.com peerURLs=https://*192.168.55.8*:2380 clientURLs=https://192.168.55.8:2379 isLeader=true

Add more members

In the instance joining the cluster:

  1. Get the etcd name for the instance in the ETCD_NAME variable:

    # grep ETCD_NAME /etc/etcd/etcd.conf
  2. Get the IP where etcd listens for peer communication:

    # grep ETCD_INITIAL_ADVERTISE_PEER_URLS /etc/etcd/etcd.conf
  3. Delete the previous etcd data:

    # rm -Rf /var/lib/etcd/*
  4. On the etcd host where etcd is properly running, add the new member:

    $ etcdctl2 member add <name> <advertise_peer_urls>
  5. The command outputs some variables. For example:

    ETCD_NAME="master2"
    ETCD_INITIAL_CLUSTER="master1=https://10.0.0.7:2380,master2=https://10.0.0.5:2380"
    ETCD_INITIAL_CLUSTER_STATE="existing"

    Add those values to the /etc/etcd/etcd.conf file of the new host:

    # vi /etc/etc/etcd.conf
  6. Once those values are replaced, start the etcd service in the node joining the cluster:

    # systemctl start etcd.service
  7. Check for error messages:

    $ journalctl -fu etcd.service
  8. Repeat the above for every etcd node joining the cluster.

  9. Verify the cluster status and cluster health once all the nodes joined:

    # etcdctl2 member list
    5cd050b4d701: name=master1 peerURLs=https://10.0.0.7:2380 clientURLs=https://10.0.0.7:2379 isLeader=true
    d0c57659d8990cbd: name=master2 peerURLs=https://10.0.0.5:2380 clientURLs=https://10.0.0.5:2379 isLeader=false
    e4696d637de3eb2d: name=master3 peerURLs=https://10.0.0.6:2380 clientURLs=https://10.0.0.6:2379 isLeader=false
    # etcdctl2 cluster-health
    member 5cd050b4d701 is healthy: got healthy result from https://10.0.0.7:2379
    member d0c57659d8990cbd is healthy: got healthy result from https://10.0.0.5:2379
    member e4696d637de3eb2d is healthy: got healthy result from https://10.0.0.6:2379
    cluster is healthy

Restoring etcd for v3

The restore procedure for v3 data is similar to the v2 data.

Snapshot integrity may be optionally verified at restore time. If the snapshot is taken with etcdctl snapshot save, it will have an integrity hash that is checked by etcdctl snapshot restore. If the snapshot is copied from the data directory, there is no integrity hash and it will only restore by using --skip-hash-check.

The procedure to restore only the v3 data must be performed on a single etcd host. You can then add the rest of the nodes to the cluster.

Procedure

  1. Stop all etcd services:

    # systemctl stop etcd.service
  2. Clear all old data, because etcdctl recreates it in the node where the restore procedure is going to be performed:

    # rm -Rf /var/lib/etcd
  3. Use the snapshot restore command with the data from /etc/etcd/etcd.conf to match the following command:

    # etcdctl3 snapshot restore */backup/etcd-xxxxxx/backup.db* \
      --data-dir /var/lib/etcd \
      --name *master-0.example.com* \
      --initial-cluster *"master-0.example.com=https://192.168.55.8:2380"* \ --initial-cluster-token *"etcd-cluster-1"* \
      --initial-advertise-peer-urls *https://192.168.55.8:2380*
    
    2017-10-03 08:55:32.440779 I | mvcc: restore compact to 1041269
    2017-10-03 08:55:32.468244 I | etcdserver/membership: added member 40bef1f6c79b3163 [https://192.168.55.8:2380] to cluster 26841ebcf610583c
  4. Restore permissions and selinux context to the restored files:

    # chown -R etcd.etcd /var/lib/etcd/
    # restorecon -Rv /var/lib/etcd
  5. Start the etcd service:

    # systemctl start etcd
  6. Check for any error messages:

    $ journalctl -fu etcd.service

Adding more nodes

Once the first instance is running, it is safe to restore multiple etcd servers as desired.

Procedure

  1. Get the etcd name for the instance in the ETCD_NAME variable:

    # grep ETCD_NAME /etc/etcd/etcd.conf
  2. Get the IP where etcd listens for peer communication:

    # grep ETCD_INITIAL_ADVERTISE_PEER_URLS /etc/etcd/etcd.conf
  3. On the etcd host where etcd is still running, add the new member:

    # etcdctl3 member add *<name>* \
      --peer-urls="*<advertise_peer_urls>*"
  4. The command outputs some variables. For example:

    ETCD_NAME="master2"
    ETCD_INITIAL_CLUSTER="master-0.example.com=https://192.168.55.8:2380"
    ETCD_INITIAL_CLUSTER_STATE="existing"

    Add those values to the /etc/etcd/etcd.conf file of the new host:

    # vi /etc/etc/etcd.conf
  5. In the recently added etcd node, clean the etcd data directories to ensure the proper backup is restored keeping the running copy:

    # mv /var/lib/etcd /var/lib/etcd.old
    # mkdir /var/lib/etcd
    # chown -R etcd.etcd /var/lib/etcd/
    # restorecon -Rv /var/lib/etcd/

    or wipe the etcd data directory:

    # rm -Rf /var/lib/etcd/*
  6. Start the etcd service in the recently added etcd host:

    # systemctl start etcd
  7. Check for errors:

    # journalctl -fu etcd.service
  8. Repeat the previous steps for every etcd node that is required to be added.

  9. Verify the cluster has been properly set:

    # etcdctl3 endpoint health
    https://master-0.example.com:2379 is healthy: successfully committed proposal: took = 1.423459ms
    https://master-1.example.com:2379 is healthy: successfully committed proposal: took = 1.767481ms
    https://master-2.example.com:2379 is healthy: successfully committed proposal: took = 1.599694ms
    
    # etcdctl3 endpoint status
    https://master-0.example.com:2379, 40bef1f6c79b3163, 3.2.5, 28 MB, true, 9, 2878
    https://master-1.example.com:2379, 1ea57201a3ff620a, 3.2.5, 28 MB, false, 9, 2878
    https://master-2.example.com:2379, 59229711e4bc65c8, 3.2.5, 28 MB, false, 9, 2878

Scaling etcd

Scaling the etcd cluster can be performed vertically by adding more resources to the etcd hosts, or horizontally by adding more etcd hosts.

If etcd is collocated on master instances, horizontally scaling etcd prevents the API and controller services competing with etcd for resources.

Due to the voting system etcd uses, the cluster must always contain an odd number of members.

The new host requires a fresh RHEL7 dedicated host. The etcd storage should be located on an SSD disk to achieve maximum performance and ideally on a dedicated disk mounted in /var/lib/etcd.

OpenShift Origin version 3.7 ships with an automated way to add a new etcd host using Ansible.

Prerequisites

  1. Before adding a new etcd host, perform a backup of both etcd configuration and data to prevent data loss.

  2. Check the current etcd cluster status to avoid adding new hosts to an unhealthy cluster:

    # etcdctl --cert-file=/etc/etcd/peer.crt \
              --key-file=/etc/etcd/peer.key \
              --ca-file=/etc/etcd/ca.crt \
              --peers="https://*master-0.example.com*:2379,\
              https://*master-1.example.com*:2379,\
              https://*master-2.example.com*:2379"\
              cluster-health
    member 5ee217d19001 is healthy: got healthy result from https://192.168.55.12:2379
    member 2a529ba1840722c0 is healthy: got healthy result from https://192.168.55.8:2379
    member ed4f0efd277d7599 is healthy: got healthy result from https://192.168.55.13:2379
    cluster is healthy

    Or, using etcd v3 API:

    # ETCDCTL_API=3 etcdctl --cert="/etc/etcd/peer.crt" \
              --key=/etc/etcd/peer.key \
              --cacert="/etc/etcd/ca.crt" \
              --endpoints="https://*master-0.example.com*:2379,\
                https://*master-1.example.com*:2379,\
                https://*master-2.example.com*:2379"
                endpoint health
    https://master-0.example.com:2379 is healthy: successfully committed proposal: took = 5.011358ms
    https://master-1.example.com:2379 is healthy: successfully committed proposal: took = 1.305173ms
    https://master-2.example.com:2379 is healthy: successfully committed proposal: took = 1.388772ms
  3. Before running the scaleup playbook, ensure the new host is registered to the proper Red Hat software channels:

    # subscription-manager register \
        --username=*<username>* --password=*<password>*
    # subscription-manager attach --pool=*<poolid>*
    # subscription-manager repos --disable="*"
    # subscription-manager repos \
        --enable=rhel-7-server-rpms \
        --enable=rhel-7-server-extras-rpms

    etcd is hosted in the rhel-7-server-extras-rpms software channel.

Adding a new etcd host using Ansible

Procedure

  1. Modify the Ansible inventory file and create a new group named [new_etcd] and add the new host. Then, add the new_etcd group as a child of the [OSEv3] group:

    [OSEv3:children]
    masters
    nodes
    etcd
    <new_etcd>
    
    ... [OUTPUT ABBREVIATED] ...
    
    [etcd]
    master-0.example.com
    master-1.example.com
    master-2.example.com
    
    [new_etcd]
    etcd0.example.com
  2. Run the etcd scaleup playbook from the host that executed the initial installation and where the Ansible inventory file is:

    $ ansible-playbook  /usr/share/ansible/openshift-ansible/playbooks/byo/openshift-etcd/scaleup.yml
  3. Once the above has finished, modify the inventory file to reflect the current status by moving the new etcd host from the [new_etcd] group to the [etcd] group:

    [OSEv3:children]
    masters
    nodes
    etcd
    <new_etcd>
    
    ... [OUTPUT ABBREVIATED] ...
    
    [etcd]
    master-0.example.com
    master-1.example.com
    master-2.example.com
    etcd0.example.com
  4. If using Flannel, modify the flanneld service configuration, located at /etc/sysconfig/flanneld on every OpenShift Origin host, to include the new etcd host:

    FLANNEL_ETCD_ENDPOINTS=https://master-0.example.com:2379,https://master-1.example.com:2379,https://master-2.example.com:2379,*https://etcd0.example.com:2379*
  5. Restart the flanneld service:

    # systemctl restart flanneld.service

Manually adding a new etcd host

The following steps can be performed on any etcd member. If using the Ansible installer, the first host provided in the [etcd] Ansible inventory is used to generate the etcd configuration and certificates stored in /etc/etcd/generated_certs, so perform the next steps in that etcd host.

Steps to be performed on the current etcd cluster

Procedure

  1. In order to create the etcd certificates, run the openssl command with the proper values. To make this process easier, create some environment variables:

    export NEW_ETCD_HOSTNAME="*etcd0.example.com*"
    export NEW_ETCD_IP="*192.168.55.21*"
    
    export CN=$NEW_ETCD_HOSTNAME
    export SAN="IP:${NEW_ETCD_IP}"
    export PREFIX="/etc/etcd/generated_certs/etcd-$CN/"
    export OPENSSLCFG="/etc/etcd/ca/openssl.cnf"

    The custom openssl extensions used as etcd_v3_ca_* include the $SAN environment variable as subjectAltName. See /etc/etcd/ca/openssl.cnf for more information.

  2. Create the directory where the configuration and certificates are stored:

    # mkdir -p ${PREFIX}
  3. Create the server certificate request and sign it:

    # openssl req -new -config ${OPENSSLCFG} \
        -keyout ${PREFIX}server.key  \
        -out ${PREFIX}server.csr \
        -reqexts etcd_v3_req -batch -nodes \
        -subj /CN=$CN
    
    # openssl ca -name etcd_ca -config ${OPENSSLCFG} \
        -out ${PREFIX}server.crt \
        -in ${PREFIX}server.csr \
        -extensions etcd_v3_ca_server -batch
  4. Create the peer certificate request and sign it:

    # openssl req -new -config ${OPENSSLCFG} \
        -keyout ${PREFIX}peer.key \
        -out ${PREFIX}peer.csr \
        -reqexts etcd_v3_req -batch -nodes \
        -subj /CN=$CN
    
    # openssl ca -name etcd_ca -config ${OPENSSLCFG} \
      -out ${PREFIX}peer.crt \
      -in ${PREFIX}peer.csr \
      -extensions etcd_v3_ca_peer -batch
  5. Copy the current etcd configuration and ca.crt files from the current node as examples to be modified later:

    # cp /etc/etcd/etcd.conf ${PREFIX}
    # cp /etc/etcd/ca.crt ${PREFIX}
  6. Add the new host to the etcd cluster. Note the new host is not configured yet so the status stays as unstarted until the new host is properly configured:

    # etcdctl2 member add ${NEW_ETCD_HOSTNAME} https://${NEW_ETCD_IP}:2380

    This command outputs the following variables:

    ETCD_NAME="<NEW_ETCD_HOSTNAME>"
    ETCD_INITIAL_CLUSTER="<NEW_ETCD_HOSTNAME>=https://<NEW_HOST_IP>:2380,<CLUSTERMEMBER1_NAME>=https:/<CLUSTERMEMBER2_IP>:2380,<CLUSTERMEMBER2_NAME>=https:/<CLUSTERMEMBER2_IP>:2380,<CLUSTERMEMBER3_NAME>=https:/<CLUSTERMEMBER3_IP>:2380"
    ETCD_INITIAL_CLUSTER_STATE="existing"
  7. Those values must be overwritten by the current ones in the sample ${PREFIX}/etcd.conf file. Also, modify the following variables with the new host IP (${NEW_ETCD_IP} can be used) in that file:

    ETCD_LISTEN_PEER_URLS
    ETCD_LISTEN_CLIENT_URLS
    ETCD_INITIAL_ADVERTISE_PEER_URLS
    ETCD_ADVERTISE_CLIENT_URLS
  8. Modify the ${PREFIX}/etcd.conf file and check for syntax errors or missing IPs otherwise the etcd service could fail:

    # vi ${PREFIX}/etcd.conf
  9. Once the file has been properly modified, a tgz file with the certificates, the sample configuration file, and the ca is created and copied to the new host:

    # tar -czvf /etc/etcd/generated_certs/${CN}.tgz -C ${PREFIX} .
    # scp /etc/etcd/generated_certs/${CN}.tgz ${CN}:/tmp/

Steps to be performed on the new etcd host

The new host is required to be subscribed to the proper Red Hat software channels as explained above in the prerequisites section.

Procedure

  1. Install iptables-services to provide iptables utilities to open the required ports for etcd:

    # yum install -y iptables-services
  2. Create firewall rules to allow etcd to communicate:

    • Port 2379/tcp for clients

    • Port 2380/tcp for peer communication

      # systemctl enable iptables.service --now
      # iptables -N OS_FIREWALL_ALLOW
      # iptables -t filter -I INPUT -j OS_FIREWALL_ALLOW
      # iptables -A OS_FIREWALL_ALLOW -p tcp -m state --state NEW -m tcp --dport 2379 -j ACCEPT
      # iptables -A OS_FIREWALL_ALLOW -p tcp -m state --state NEW -m tcp --dport 2380 -j ACCEPT
      # iptables-save | tee /etc/sysconfig/iptables

      In this example, a new chain OS_FIREWALL_ALLOW is created, which is the standard naming the OpenShift Origin installer uses for firewall rules.

      If the environment is hosted in an IaaS environment, modify the security groups for the instance to allow incoming traffic to those ports as well.

  3. Install etcd software:

    # yum install -y etcd
  4. Ensure the service is not running:

    # systemctl disable etcd --now
  5. Remove any etcd configuration and data:

    # rm -Rf /etc/etcd/*
    # rm -Rf /var/lib/etcd/*
  6. Untar the certificates and configuration files

    # tar xzvf /tmp/*etcd0.example.com*.tgz -C /etc/etcd/
  7. Restore etcd configuration and data owner:

    # chown -R etcd.etcd /etc/etcd/
    # chown -R etcd.etcd /var/lib/etcd/
  8. Start etcd on the new host:

    # systemctl enable etcd --now
  9. Verify the host has been added to the cluster and the current cluster health:

    # etcdctl --cert-file=/etc/etcd/peer.crt \
              --key-file=/etc/etcd/peer.key \
              --ca-file=/etc/etcd/ca.crt \
              --peers="https://*master-0.example.com*:2379,\
              https://*master-1.example.com*:2379,\
              https://*master-2.example.com*:2379,\
              https://*etcd0.example.com*:2379"\
              cluster-health
    member 5ee217d19001 is healthy: got healthy result from https://192.168.55.12:2379
    member 2a529ba1840722c0 is healthy: got healthy result from https://192.168.55.8:2379
    member 8b8904727bf526a5 is healthy: got healthy result from https://192.168.55.21:2379
    member ed4f0efd277d7599 is healthy: got healthy result from https://192.168.55.13:2379
    cluster is healthy

    Or, using etcd v3 API:

    # ETCDCTL_API=3 etcdctl --cert="/etc/etcd/peer.crt" \
              --key=/etc/etcd/peer.key \
              --cacert="/etc/etcd/ca.crt" \
              --endpoints="https://*master-0.example.com*:2379,\
                https://*master-1.example.com*:2379,\
                https://*master-2.example.com*:2379,\
                https://*etcd0.example.com*:2379"\
                endpoint health
    https://master-0.example.com:2379 is healthy: successfully committed proposal: took = 5.011358ms
    https://master-1.example.com:2379 is healthy: successfully committed proposal: took = 1.305173ms
    https://master-2.example.com:2379 is healthy: successfully committed proposal: took = 1.388772ms
    https://etcd0.example.com:2379 is healthy: successfully committed proposal: took = 1.498829ms

Steps to be performed on all OpenShift Origin masters

Procedure

  1. Modify the master configuration to add the new etcd host to the list of the etcd servers OpenShift Origin uses to store the data, located in the etcClientInfo section of the /etc/origin/master/master-config.yaml file on every master:

    etcdClientInfo:
      ca: master.etcd-ca.crt
      certFile: master.etcd-client.crt
      keyFile: master.etcd-client.key
      urls:
        - https://master-0.example.com:2379
        - https://master-1.example.com:2379
        - https://master-2.example.com:2379
        *- https://etcd0.example.com:2379*
  2. Restart the master API service on every master:

    # systemctl restart atomic-openshift-master-api

    Or, on a single master cluster installation

    # systemctl restart atomic-openshift-master

The number of etcd nodes must be odd, so at least two hosts must be added.

  1. If using Flannel, the flanneld service configuration located at /etc/sysconfig/flanneld on every OpenShift Origin host must be modified to include the new etcd host:

    FLANNEL_ETCD_ENDPOINTS=https://master-0.example.com:2379,https://master-1.example.com:2379,https://master-2.example.com:2379,*https://etcd0.example.com:2379*
  2. Restart the flanneld service:

    # systemctl restart flanneld.service

Removing an etcd host

An etcd host can fail beyond restoration. This section walks through removing the failed etcd host from the cluster.

Ensure the etcd cluster maintains quorum while removing the etcd host, by removing a single host at a time from a cluster.

Steps to be performed on all masters hosts

Procedure

  1. Edit the failed etcd host out of the /etc/origin/master/master-config.yaml master configuration file on every master:

    etcdClientInfo:
      ca: master.etcd-ca.crt
      certFile: master.etcd-client.crt
      keyFile: master.etcd-client.key
      urls:
        - https://master-0.example.com:2379
        - https://master-1.example.com:2379
        *- https://master-2.example.com:2379* (1)
    1 The host to be removed.
  2. Restart the master API service on every master:

    # systemctl restart atomic-openshift-master-api

    Or, if using a single master cluster installation:

    # systemctl restart atomic-openshift-master

Steps to be performed in the current etcd cluster

Procedure

  1. Remove the failed host from the cluster by running the following on a functioning etcd host:

    # etcdctl2 cluster-health
    member 5ee217d19001 is healthy: got healthy result from https://192.168.55.12:2379
    member 2a529ba1840722c0 is healthy: got healthy result from https://192.168.55.8:2379
    failed to check the health of member 8372784203e11288 on https://192.168.55.21:2379: Get https://192.168.55.21:2379/health: dial tcp 192.168.55.21:2379: getsockopt: connection refused
    member 8372784203e11288 is unreachable: [https://192.168.55.21:2379] are all unreachable
    member ed4f0efd277d7599 is healthy: got healthy result from https://192.168.55.13:2379
    cluster is healthy
    
    # etcdctl2 member remove 8372784203e11288
    Removed member 8372784203e11288 from cluster
    
    # etcdctl2 cluster-health
    member 5ee217d19001 is healthy: got healthy result from https://192.168.55.12:2379
    member 2a529ba1840722c0 is healthy: got healthy result from https://192.168.55.8:2379
    member ed4f0efd277d7599 is healthy: got healthy result from https://192.168.55.13:2379
    cluster is healthy

    The remove command requires the etcd ID, not the hostname.

  2. To ensure the etcd configuration does not use the failed host when the etcd service is restarted, modify the /etc/etcd/etcd.conf file on all remaining etcd hosts and remove the failed host in the value for the ETCD_INITIAL_CLUSTER variable:

    # vi /etc/etcd/etcd.conf

    For example:

    ETCD_INITIAL_CLUSTER=master-0.example.com=https://192.168.55.8:2380,master-1.example.com=https://192.168.55.12:2380,master-2.example.com=https://192.168.55.13:2380

    becomes:

    ETCD_INITIAL_CLUSTER=master-0.example.com=https://192.168.55.8:2380,master-1.example.com=https://192.168.55.12:2380

    Restarting the etcd services is not required, because the failed host has been removed using etcdctl on the command line.

  3. Modify the Ansible inventory file to reflect the current status of the cluster and to avoid issues if running a playbook:

    [OSEv3:children]
    masters
    nodes
    etcd
    
    ... [OUTPUT ABBREVIATED] ...
    
    [etcd]
    master-0.example.com
    master-1.example.com
  4. If using Flannel, modify the flanneld service configuration located at /etc/sysconfig/flanneld on every host and remove the etcd host:

    FLANNEL_ETCD_ENDPOINTS=https://master-0.example.com:2379,https://master-1.example.com:2379,https://master-2.example.com:2379
  5. Restart the flanneld service:

    # systemctl restart flanneld.service

Replacing an etcd host

To replace an etcd host, first remove the etcd node from the cluster following the steps from Removing an etcd host, then scale up the etcd cluster with the new host using the scale up Ansible playbook or the manual procedure in Scaling etcd.

The etcd cluster should maintain a quorum during the replacement operation. This means that at least one host should be in operation at all times.

If the host replacement operation occurs while the etcd cluster maintains a quorum, cluster operations are not affected, except if there is a large etcd data to replicate where some operations can be slowed down.

Ensure a backup of etcd data and configuration files exists before any procedure involving the etcd cluster to ensure restoration in the case of failure.