RDF/IMP and IMPX System Management Manual (RDF 1.3+)

Managing RDF
Compaq NonStop™ RDF/IMP and IMPX System Management Manual—522204-001
5-9
TMF Subsystem Crash
TMF Subsystem Crash
RDF can recover from a TMF crash on the primary system if the TMF volume recovery
operation is successful after the crash. For this recovery, proceed as follows:
1. Stop RDF on the primary system by entering the following command through
RDFCOM:
]STOP RDF
2. Restart TMF by entering the following command sequence through TMFCOM:
~DISABLE DATAVOLS *
~START TMF
Notice that these commands prevent any disk volumes on the local system from
being enabled for TMF operations before starting the subsystem.
3. Reenable all pertinent disk volumes for TMF operations by entering the following
command through TMFCOM:
~ENABLE DATAVOLS *
When this command is executed, TMF performs its volume recovery operation on
the audited volumes, and RDF reads the audit during this operation.
4. Restart your applications.
5. Restart RDF through RDFCOM by entering:
]START RDF
RDF can recover from a TMF crash on the backup system if the TMF volume recovery
operation is successful after the crash. For this recovery, proceed as follows:
1. From the primary system, stop updating of the backup database by entering this
command through RDFCOM:
]STOP UPDATE
2. From the backup system, stop TMF on the backup system by entering the following
command through TMFCOM:
~STOP TMF
3. Correct the problem on the backup system and recover the volume.
Note. Normally you start RDF before starting your applications. If the TMF subsystem has
crashed on the primary system, however, characteristics of the TMF audit trail after the TMF
subsystem is restarted require that you restart your applications before restarting RDF. This
allows audit information to be generated in the MAT and any auxiliary audit trails involved in
your RDF environment. At that point you restart RDF.
If you started RDF before starting your applications, you might notice that the auxiliary
extractor continues to read the same data over and over again without moving forward. To
correct that situation, you merely stop and then restart the RDF subsystem.