IV71572: SHUTDOWN -F ON POWERHA MAY PANIC INSTEAD OF HALT

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• Running shutdown -f in PowerHA may result in a system panic
  instead of a halt, due to the presence of a new RSCT Dead Man
  Switch, owned by the Group Services subsystem, to better
  protect RSCT integrity in the CAA environment.
  
  Customers at rsct.basic.rte 3.1.5.6 (or higher) or
  rsct.basic.rte 3.2.0.2 (or higher) should begin seeing the
  hags-owned DMS active; check the end of "lssrc -ls cthags"
  output to verify.
  

Local fix

• To address the issue without a code update, simply add or edit
  this file:   /etc/rc.d/rc2.d/Kcluster
  
  The content should be:
  
  # -- BEGIN Kcluster content --
  case "$1" in
  start ) exit 0
          ;;
  stop ) CLNAME=$(/usr/sbin/rsct/bin/ct_clusterinfo -c)
         if [[ -n $CLNAME && $CLNAME != "IW" ]]; then
            /usr/sbin/rsct/bin/hags_disable_client_kill -s cthags
            /usr/sbin/rsct/bin/hags_stopdms -s cthags
            stopsrc -s IBM.ConfigRM
            exit 0
         fi
         ;;
  * ) echo "Usage: $0 (start | stop)"
      exit 1
  esac
  exit 0
  # -- END Kcluster content --
  
  If the Kcluster file already exists, it will have some of that
  content (plus extra comments), without the "hags_stopdms" line.
  Just add that missing line in the sequence shown above.
  
  Note: For customers in need of HIPER APARs IV66606 or IV69760,
  this fix for /etc/rc.d/rc2.d/Kcluster IS included in those
  downloadable ifixes.
  

Problem summary

• A solution for preventing RSCT-layer critical reactions in
  the middle of a shutdown of a PowerHA node was provided by
  PowerHA in an rc.d script named "Kcluster".
  This script was missing a recently-introduced HAGS DMS
  enablement factor, and it was further decided that trying
  to handle updates or variations in RSCT protection methods
  would not be practical through a PowerHA deliverable.
  

Problem conclusion

• New RSCT-owned rc.d scripts are being introduced with this
  APAR, which will handle any AIX shutdown needs on a
  PowerHA-installed node.  It will not conflict with the
  existing Kcluster file, if already installed.  Note that
  the new RSCT rc.d scripts are specific to PowerHA, and are
  not applicable to other domain types which might make use
  of RSCT clustering services.
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

  IV87543 IV87544

Fix information

• Fixed component name

  RSCT/RMC FOR CS

• Fixed component ID

  5765F07AP

Applicable component levels

• R315 PSY U874795

     UP16/10/20 I 1000

PTF to Fileset Mapping

• U874795 rsct.core.utils 3.1.5.13

• U874795 rsct.core.utils 3.1.5.13

• U874795 rsct.core.utils 3.1.5.13