IT21112: CLUSTER SENDER CHANNELS FAIL TO RESTART FOLLOWING CPU FAILURE ON 2 CPU SYSTEM

WebSphere MQ for HP NonStop Server V5.3.1 fix pack 5.3.1.14

APAR status

• Closed as program error.

Error description

• The issue ocurs only when the primary instances of the channel
  server and the execution controller are on the failing CPU.
  Root cause is a race condition during takeover between
  MQCHANSVR and MQECSVR, when no new backup process could be
  created. Usually this happens only during CPU down on a two CPU
  system
  

Local fix

• Once the second cpu is back up and the backup processes are back
  running, the issue situation, if present, could be resolved, by
  1. manually stopping the primary MQCHSVR using its CPU,PID
  2. manually stopping MCA, PID shown in channelstatus
   mqsc> display chs(CHANNEL) all
  Additionally, the problem could be avoided on 2 cpu systems, by
  placing primary MQCHSVR on the CPU running the backup MQECSVR.
  

Problem summary

• The root cause is a race condition between failover procedures
  of the MQECSVR and the MQCHSVR, when both primaries fail
  simultaneously and no new backup processes could be created.
  This usually only happens on CPU down on two CPU systems.
  Due to the race condition the MQCHSVR uses the process-handle
  of the failed/lost primary MQECSVR during EC health check,
  resulting in initialization problems, and ending up with
  this reported issue.
  

Problem conclusion

• Added additional step to EC health check, retrieve current
  process-handle for known process name of MQECSVR using OS APIs,
  before performing health check operations requiring
  process-handle of MQECSVR.
  

Temporary fix

• Configure MQ pathway environment on two CPU systems, to start
  primary MQCHSVR on the CPU hosting the backup of MQECSVR.
  This avoids the problem at least for one CPU down.
  

Comments

APAR Information

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

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

Fix information

• Fixed component name

  WEBS MQ NSS ITA

• Fixed component ID

  5724A3902

Applicable component levels

• R531 PSY

     UP