Meet the software requirements

The following prerequisites must be met for the installation:

• Make sure that the nodes and NFS server for the installation meet the minimum system requirements. For details, see Meet the hardware requirements.

• The master and worker nodes must have a static IP address.

• The host names of the master and the worker nodes must be DNS resolvable (not only via /etc/hosts). Alternatively, it is also possible to resolve the host names / IP addresses via the following local hosts files:

  /etc/hosts

  /var/vols/itom/core/baseinfra-1.0/kube-dns-hosts/hosts

• The /tmp directory of the (first) master node must have at least 2.5 GB of space available when adding worker nodes from the management portal.

• If the machine already has Docker or Kubernetes installed, uninstall them.

• Make sure you configured your firewall to allow the necessary ports. For details, see Enable firewall on a running node.

• The following ports are needed on all nodes during and after the installation, and should not be used by another application: 383, 443, 2380, 4001, 4243, 5000, 5432, 5443, 8080, 8200, 8201, 8443, 10249, 10250, 10251, 10252, 10255, 31387, 31389.

  The following ports must be open for system processes: 111 (rpcbind), 2049 (NFS), 20048 (rpc.mountd).

  Note The installation script checks and reports if necessary ports are in use.

• Check if you have installed the following rpm packages on all nodes:

  rpm -qa | grep -E "java-1.8.0-openjdk|libgcrypt|libseccomp|libtool-ltdl|net-tools|nfs-utils|systemd-libs|device-mapper-libs|lsof|unzip|chrony|rpcbind"

  Note systemd-libs must be version 219 or higher.

  If one or multiple of the packages are not installed, install them using yum install:

  yum install java-1.8.0-openjdk libgcrypt libseccomp libtool-ltdl net-tools nfs-utils systemd-libs device-mapper-libs lsof unzip chrony rpcbind

  If you installed Chrony, run the following commands afterwards:

  systemctl start chronyd
systemctl enable chronyd

• Remove the shared NFS folder if you have previously installed the Container Deployment Foundation. The default folder is /var/vols/itom/core.
  For example: rm -rf /var/vols/itom/core/*

  Also remove the directory on the NFS server where you stored Operations Bridge suite data, if you previously installed the Operations Bridge Suite, for example:

  rm -rf /var/vols/itom/opsbridge/*

• The NFS server, the master nodes, and the worker nodes must be installed under the same subnet.

• Make sure that the browser cache is cleared.

• The time on all master and worker nodes must be the same. To synchronize the time on your nodes, you can, for example, use NTP or VMWare tools.

• For all processes in your /etc/ environment, make sure that https_proxy and http_proxy settings are not set (unset https_proxy; unset http_proxy). Alternatively, add the IP address of the master node to the no_proxy list for all master and worker nodes.

  Example: export no_proxy="localhost,127.0.0.1,<master_node_IP>"

  In a multiple master node deployment, add the IP address you specified in HA_VIRTUAL_IP (virtual IP shared by multiple master nodes) to the no_proxy list.
  Make sure that these settings are consistent on all master and worker nodes.