Rebuilding an index

Rebuild an index if it is corrupted.

Before you begin

Test the rebuild index procedure on a test cluster.

About this task

For the rebuilding instance, disable all incoming interfaces (MCI, BDI, FLI, API) during the rebuild. Do not change the interface or cluster during the rebuild.

Run the index rebuild on only one instance and restore the other instances later from this instance after you verify that the rebuild was successful. If deferred writing is enabled on that instance, the rebuild index function normally finishes relatively quickly.

Procedure

  1. Switch the API to an instance that is not rebuilding the index. It is not possible to switch the API after rebuild is started.
  2. Ensure that a user has maintenance privilege to rebuild the index.
  3. Enable log message 0. It prints log messages that help to estimate the rebuild duration after it is started.
  4. Make sure that FLI to the rebuilding instance is synchronized.
  5. Disable incoming interfaces (MCI, BDI, FLI, API) on the rebuilding instance.
  6. Disable log message 0 again if the instance with active API is not the rebuilding instance.
  7. Click Cluster > Maintenance > rebuild index.
  8. Wait for the log message that indicates that the index rebuild completed successfully.
  9. Enable incoming interfaces.