The procedure for manually upgrading cluster nodes is similar to upgrading a standalone Carbon Black EDR server.

See Upgrading a Server. Additional steps are required to ensure that the cluster nodes are communicating with each other.

To manually upgrade minion servers in a cluster:

  1. After the primary node has been upgraded, login to each of the other machines in the cluster and upgrade cb-enterprise:
    yum upgrade cb-enterprise
  2. Login to each of the cluster nodes and upgrade their data schema to the latest version:
    /usr/share/cb/cbupgrade [--proceed-on-rpmnew] [--non-interactive]

    With no argument, this command stops running if Carbon Black EDR RPMNEW files are detected, indicating a modified configuration file. The command reports the names of those files. You must manually resolve differences between the RPMNEW and the existing configuration file of the same name, and delete the RPMNEW file before cbupgrade completes.

    With the proceed-on-rpmnew argument, cbupgrade runs to completion but also reports the RPMNEW files that it discovered so that you can address them after the upgrade. This can be helpful for a seamless upgrade and are comfortable editing the configuration files later.

    When the --proceed-on-rpmnew option is used with --non-interactive option, you are not prompted for the upgrade to proceed, but the cb-enterprise services do not start automatically as if you ran the command without the non-interactive option.

    Caution: At the end of the cbupgrade process, if you are prompted to start services, do not start the services.
  3. With a new software version, communications port requirements can change. You must check whether the firewall settings require an update.
    1. If you are manually updating the settings, run the following command to identify which rules need to be added:
      /usr/share/cb/cbcheck firewall -l
    2. To adjust the firewall settings automatically, run this command to have the utility apply those settings:
      usr/share/cb/cbcheck firewall --apply
  4. After all nodes have been upgraded (but not started), start the entire cluster by logging into the primary node and issuing the following command:
    /usr/share/cb/cbcluster start

    This is an important step to perform after an upgrade. When the cluster is started using the cbcluster tool, it redistributes configuration files that must be synchronized across all nodes in the cluster. If a new software version introduces configuration changes, this step ensures that each minion node has the updates.