RSC/MP 7.2 Programming Manual

Message Handling
HP NonStop Remote Server Call (RSC/MP) Programming Manual—522360-004
7-7
Recovering the TDP After a Failure
and the primary destination. After receipt of a connection failure error, set the
HOST_PIPENAME option to the alternate destination and reissue an RscConnect.
To reduce the possibility of connection failure errors, multiple NIFs can be configured
on the client workstation. The alternate NIF(s) could be configured to use alternate
LAN segments or serial communications lines. In the case of the failure of a
communication path, Pipeman attempts to reach the destination over all configured
NIFs before declaring a path failure. See the Remote Server Call (RSC/MP) Installation
and Configuration Guide for more information.
Any transactions that are active at the time of a failure are automatically aborted and
must be restarted. Any pending nowaited I/Os or UMS messages are discarded.
Recovering the TDP After a Failure
If a network failure occurs, the object in the TDP for that communication system is
aborted. If the AUTORESTART attribute is set for that object, the TDP automatically
attempts to connect and resend the aborted transactions. For example, if the CPU in
which the Pipeman process is running goes down, the PIPE object is aborted. The
TDP aborts all TMF transactions for all terminal sessions communicating through that
PIPE object (see Monitoring Transaction Calls on page 6-6). The API returns an error
from the RSC_PIPE_SUBSYSTEM to the workstation.
The TDP on the NonStop system continues to check if the Pipeman process for the
PIPE object is available. If the AUTORESTART attribute for the PIPE object is set as a
nonzero value after the backup Pipeman process takeover is complete, the TDP can
communicate with the workstation again. The workstation must connect again and
resend the transaction to the TDP.
The default value for the AUTORESTART attribute is 30 seconds. Use RSCCOM to
configure AUTORESTART.
If the TDP process terminates abnormally, the NonStop host attempts to switch to a
backup process. If you use TMF, data integrity is maintained because the NonStop
Kernel forces any open transactions belonging to a process that terminated to be rolled
back. See Persistent Applications on page 6-15, for more information about persistent
applications and running the TDP in persistent mode.