A hot I/O condition occurs when a device, control unit, or channel
path causes continuous unsolicited I/O interrupts. The operating system
attempts to recover from a hot I/O condition so that a reIPL is not
required. For diagnostic purposes, the operating system indicates
all hot I/O incidents in logrec data set error records.
The operating system first tries recovery at the device level by
issuing the Clear Subchannel (CSCH) instruction in an attempt to clear
the hot I/O condition. If the condition is cleared, processing continues
normally. If the condition persists, the next recovery action is
determined by one of the following:
- The parameters the installation defined in the IECIOSxx parmlib
member for hot I/O recovery
- Operator response to the appropriate hot I/O message
or restartable wait state for the class of device:
- Message IOS117A, (IOS110D, or wait state X'110') for non-DASD,
non-dynamic pathing device
- Message IOS118A, (IOS111D, or wait state X'111') for DASD
or dynamic pathing device that is not reserved
- Message IOS119A, (IOS112D, or wait state X'112') for DASD
or dynamic pathing device that is reserved
Because IPLs related to hot I/O are generally caused by incorrect
operator actions, an installation should use the IECIOSxx parmlib
member to make hot I/O recovery more automatic and reduce the need
for immediate operator intervention. The following example parameters,
when defined in the IECIOSxx parmlib member, tell the operating system
how to handle automatic recovery from hot I/O.