Recovering from a Data Corruption Failure

Recovering from a Data Corruption Failure

If a disk fails, a file system becomes corrupted, or a data directory gets accidentally deleted, the segment instance serving that data becomes unoperational.

To recover with mirroring enabled

1.Troubleshoot and correct the problem that caused the failure. For example, you may need to replace the disk or rebuild the file system.

2.Run the gprecoverseg utility to recover the failed segment instances. For example:

$ gprecoverseg

3.The recovery process will shut down the system in either read-only or continue mode. The system cannot recover failed segments without shutting down.

4.When the recovery process is complete, run the gpstate utility to verify that all segment instances are up:

$ gpstate

To recover without mirroring enabled

If you do not have mirroring enabled, there is no way to recover a single segment that has corrupted or missing data. In this scenario, you will have to restore the entire database from your most recent Greenplum Database backup to ensure data integrity across all segments.