Performing a rolling upgrade (rolling restart)

You can upgrade Pega Platform by performing a rolling upgrade (rolling restart) that keeps your system always available during upgrade. You remove nodes from the load balancer, shut them down, upgrade, and restart each node one by one. You do not add them back to the load balancer until you have upgraded half of the nodes.

To perform a rolling upgrade, complete the following steps:

  1. Prepare to upgrade your database by doing the following steps.
    1. Disable rule saving. For more information, see Disabling rule creation on the rules schema in the appropriate Upgrade guide for your system.
    2. Migrate the PegaRULES schema to a temporary schema. For more information, see Migrating the existing rules schema in the appropriate Upgrade guide for your system.
    3. Update the new rules schema, for example, a framework or application update. For more information, see Upgrading the migrated rules schema in the appropriate Upgrade guide for your system.
    4. Copy the new rule schema to the production database. For more information, see Migrating to the new rules schema in the appropriate Upgrade guide for your system.
  2. Optional: To switch to Apache Ignite client-server mode, start the stand-alone Apache Ignite servers before upgrading the nodes. For more information, see Apache Ignite client-server clustering topology and Deploying and starting the Apache Ignite servers.
  3. Upgrade half the nodes one by one.
    1. Configure the load balancer to disable a node.
      Note:
      • Disabling the node does not allow new connections, but it allows existing users and services to complete work.
      • Quiescing a Pega Platform node that has not been disabled in the load balancer results in error conditions for users of that Pega Platform node, because new users cannot log in. The Pega Platform must be disabled in the load balancer so that new users are redirected to another active Pega Platform node.
    2. Quiesce the Pega Platform node, by using the Autonomic Event Services (AES), System Management Administrator (SMA), or high availability landing pages. For more information, see Cluster management.
    3. Ensure that all requestors are passivated and the system run state is set to "Quiesce Complete", by using the HA Cluster Management landing page.
    4. Shut down the node.
    5. Update the data source to connect to the updated schema. For more information, see Upgrading the data schema in the appropriate upgrade guide for your system.
    6. Start the node.
    7. Perform any needed post-upgrade activities and tests.
  4. After you upgrade half of nodes, disable the remaining non-upgraded nodes in the load balancer.
  5. Add all the upgraded nodes, which you upgraded in step 3, back to the load balancer to start taking traffic.
  6. Upgrade the remaining half of the nodes (the non-upgraded nodes) one by one.
    1. Perform steps 3 b through 3 g.
    2. Add the node back to the load balancer to start taking traffic.