Custodians always record the last Id of the record that they process.  In the event of a system crash or failure, a simple restart of CS:Govern and resume of the Custodian, will result in that Custodian continuing on from where it left off as though nothing ever went wrong.

Additionally, as new records are brought down by CopyStorm, which include Governed fields, those fields will be protected (encrypted and masked) as expected.

Note, however, that during the time between when the system crashed and new records are brought down from CopyStorm, there will be a certain number of records that should have been protected by the Custodian that have not (that is, assuming this is an Encryption Custodian).  There are two ways to ensure that the gap in these records is filled:

1. After a system crash and the machine has been brought back to normal functioning levels (e.g. a system reboot), the CS:Govern user can manually launch the Custodian that was running prior to the failure.
2. The Custodian runner (see Custodian Runner Tab) can be used to detect that a previous run of a Custodian was stopped midstream due to a system failure, and will automatically resume that Custodian (or Custodians.)

Of the two, clearly the second option is the preferred one and the solution most recommended, as it is an automated process that requires no manual intervention.