IBM System/370 145 Manual page 109

The failing CCW is reissued twice by the control unit. If one of
the two retries is succelE;sful, the channel program continues
normally.
3. When a seek malfunction is detected.
The control unit retries the command ten times in an attempt to
position the access mechanism correctly_
4. When an alternate or
def~~ctive
track condition is recognized
before data transfer begins.
The control unit determines toe location of the alternate or
defective track (from RO on the track), initiates a.seek to this
track, and orients to thE:! index J?oint. When this sequence
completes, the original c::ommand
1S
reissued by the control unit.
This is a programmed pro(::edure for previously announced
System7360 direct access devices.
5. When a command overrun (or late command-chaining) condition
occurs during execution of a channel program because of
interference from another channel or the CPU .•
The control unit initiatE~s a retry of the command that was late.
6. When a data overrun occw:~s. except for:
a. A data overrun that occurs during a track overflow operation
in the second or subsequent segments.
b. A data overrun that occurs during a formatting write.
Error logging. Usage and er:r:~or counters for each drive in the
facility are maintained continuously in the control unit. The usage
counters are used to accumulate the number of bytes read and seeks
issued. The error counters are used to accumulate the number of seek,
correctable data, and uncorrectalble data errors that were retried
successfully by a command retry procedure, as already described. Also
accumulated is the total number of command and data overrun conditions
that were retried by the storage control unit.
When a counter reaches its
th~eshold
or when a pack is removed from a
drive, the control unit indicate~s the condition via a unit check when
the next I/O operation is initia:J:ed to the drive. Counter data can be
obtained and counters can be reset by issuing a SENSE or READ LOG
command. These statistics can then be logged in the system error data
set for later diagnosis.
Inline diagnostiC tests. A 3830 Model 1 control unit can execute
diagnostic tests on a malfunctioning drive while normal operations take
place on the remaining drives in the string. After the service address
plug is inserted in the malfunctioning drive, diagnostics can be
executed on that drive by the customer engineer using the CE panel on
the 3830 control unit. Operationally, the drive is offline to the
control unit, and physically the drive is of.fline to the operating
system.
Online testing of 3330-series drives can be performed under OLTEP
control, as usual. Both OLT's and diagnostic programs contained in the
OLT library can be executed on a malfunctioning drive via OLTEP. The
diagnostic tests are loaded into control storage in the control unit
from the OLT library. Operationally, the 3330-series drive is online to
the control unit but is logically offline to the operating system.
A Guide to the IBM System/370 Model 145 99