Rebuilding an On-prem HA Search Service

Rebuilds are not simultaneously sent to each of the HA search services. Instead, rebuilds are accomplished using a single web application node and the resulting rebuild activities are delivered to a single search service. If the web application or the search service fails during a rebuild, the rebuild will have to be restarted from the beginning. You can control which search service is responsible for receiving rebuilds via the service directory configuration. The next three diagrams show how to accomplish a successful rebuild.

Note that if the document conversion node is running during a search index rebuild, the web application nodes may be more responsive to requests due to reduced loads, but the duration of the rebuild will be increased.
  1. Initiate a search rebuild from the Admin Console (System > Settings > Content Search > Rebuild Index). The rebuild activities will go to the search service configured in the service directory. Wait for the rebuild to complete. Here's what happens during this process:
    Diagram of Starting a Search Rebuild
  2. Shut down the ingress replicators with service jive-ingress-replicator-service stop. (Note that you must be logged in as the jive user to successfully execute this and the other commands listed here). Activities that the replicator service has not delivered will be delivered after the restart. While the ingress replicators are offline, the web application will journal activities to disk. After the ingress replicator service is restarted, these journalled activities will be delivered to the ingress replicators.
  3. Shut down search service 2 (service jive-search-service stop).
  4. Copy the active index from search service 1 to search service 2.
  5. Restart search service 2 (service jive-search-service start). This process looks something like this:
    Diagram of Copying the Search Index
  6. Restart ingress replicators 1 and 2 (service jive-ingress-replicator-service start).
    Diagram of Rebuilding a Search Index