Previous Topic: 00316: CPU Degraded by TLB/Cache Buffer Failure(s)

Next Topic: 5. FILES

00319: Recovery of I/O on Failed Processor In Progress


FILE              TSO System Activity
SAS FILE NAME     DETAIL.TSOTSO01
SOURCE LOCATION   sharedprefix.MICS.SOURCE(DYTSOEXC)

SEVERITY          Critical      (SEVERITY='C')
MANAGEMENT AREA   Performance   (MGMTAREA='PERFORMANCE')

PURPOSE      Indicates that multiprocessor is running in
             degraded mode.

RATIONALE    The 168-MP, 3033-MP, and the 3081 all support
             (optional on 168-MP) some scheme for continuing
             to do I/O to devices normally only accessible
             from a CPU that has failed.  Such devices are
             deemed to be "asymmetrically attached" because
             they lack paths (channels) from both sides of
             the MP/dyadic processors.  Any device may be
             asymmetrically attached.  Certain old-style TP
             devices (e.g., BSC lines going through 3705-EP)
             must be.

             In 168-MP, the recovery scheme is called "CRH"
             for Channel Recovery Hardware, a less
             sophisticated method of issuing DIAGNOSE
             instructions to connect and disconnect channel 6
             of the healthy processor to a given channel of
             the dead one and then issue SIOs and/or "poll"
             for I/O interrupts.  In newer processors, a more
             architecturally complex solution called "channel
             set switching" has been implemented.  In either
             case, the loss of a processor and its I/O is a
             serious matter that can manifest itself in a
             number of ways, such as TP line time outs.

DEFINITION   This exception is noted when either of the
             following data elements contains a nonzero
             value:

             TSOXCRHY - Intvls When Chan Reconfig Hard Active
             TSOXCSSY - Intvls When Chan Set Switching Active

             See Appendix B for more information on this
             data element.

EXCEPTION    The SAS statements identifying the exception
STATEMENTS   situation and describing the condition are
             stored in the source member named in SOURCE
             LOCATION and are described below:


************************************************************;
*
** 00319
** RECOVERY OF I/O ON FAILED PROCESSOR IN PROGRESS
*;
IF SUM (OF TSOXCRHY TSOXCSSY) > 0
 THEN DO;
  EXCCODE='00319'; SEVERITY='C'; MGMTAREA='PERFORMANCE';
  EXCDESC1='RECOVERY OF I/O ON FAILED PROCESSOR IN PROGRESS';
  IF TSOXCRHY > 0 THEN
       EXCDESC2='CHANL RECONFIG HARDWARE IS ACTIVE'
            || ' ENDTS=' || PUT(ENDTS,DATETIME19.2);
  IF TSOXCSSY > 0 THEN
       EXCDESC2='CHANNEL SET SWITCHING IS ACTIVE'
            || ' ENDTS=' || PUT(ENDTS,DATETIME19.2);
  LINK HIT;
END;

THRESHOLD    None required.
MODIFICATION