To restart a host, both of the following must occur:
- VTAM® representation of
the network must be restored. Using
configuration restart facilities, you can have it restored to the
state it was in before the host failed. NetView® command lists can be used to automate
the restarting of the network resources.
- Sessions must be reestablished. Some sessions can continue despite
the failure. For example, sessions between devices in this domain
and application programs in other domains that have been defined to
continue after automatic network shutdown of the NCP can continue.
Of these, some sessions also continue when VTAM is started. If the device in session supports
recovery, the session continues when VTAM reactivates physical units and LUs. If not, the session
is disrupted.
Note: If the session is established with a nonextended
BIND, the restarted VTAM loses
knowledge of all sessions. Therefore, you should plan for the following
two situations:
- Cross-domain sessions with peripheral LUs owned by the failing
host can continue when VTAM fails, but without VTAM support.
Those sessions that continue when VTAM restarts do so with limited VTAM support. That is, the DISPLAY operator command, when issued
from the restarted host, provides only limited session information.
Unformatted LOGOFF requests from these LUs must be specified as TYPE(FORCE)
or TYPE(UNCOND) and must not see the session partner; that is, the
APPLID operand must be omitted. Formatted Terminate Self requests
from these peripheral LUs must be specified with the type code set
to cleanup. TYPE=UNCOND or TYPE=FORCE must be specified in VARY TERM
operator commands.
- Sessions that do not continue after VTAM failure must be reestablished by the operator or one of
the session partners, as necessary.