IC63379: NEED TO DOCUMENT THAT THE HDR SECONDARY WILL ALWAYS BE AHEAD OF RSS IN TERMS OF REPLAYING LOGS

APAR status

• Closed as documentation error.

Error description

• The primary node ensures that a log page is not sent to any of
  RSS nodes until it has been recieved by the HDR. This is
  required since the HDR is the first choice for failover in the
  cluster.
  
  If the RSS goes ahead of HDR in terms log replay, then in the
  event of failover it will not be able to sync up with the new
  primary.
  
  If the HDR is offline, the primary will send the logs to RSS.
  But once the HDR reconnects, the primary will stop sending the
  logs to RSS till the HDR is caught up.
  
  Both these behaviors need to be documented clearly to avoid
  confusion.
  
  It could be necessary to add following in Admin Guide , under
  section 'Comparison of RS Secondary Servers and HDR Secondary
  Servers':
  "The HDR will always be the most log advanced server among the
  non-shared disk secondary nodes (RSS and HDR). This ensures that
  in the even of HDR becoming the primary, all the RSS nodes can
  continue to apply the logs from the new primary.
  If the HDR is offline, the primary server will continue to send
  logs to RSS. Once the HDR reconnects, the primary will make sure
  that its catches up and doesn't stay behind the RSS nodes."
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * All users of documentation for v11.50.xc5 and earlier.       *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * The prioritization of the HDR secondary server logs is not   *
  * clear.                                                       *
  *                                                              *
  * This omission occurs in the Comparison of RS Secondary       *
  * Servers and HDR Secondary Servers topic of the               *
  * Administrator's Guide.                                       *
  ****************************************************************
  * RECOMMENDATION:                                              *
  * Upgrade to v11.50.xc6 documentation when available.          *
  ****************************************************************
  

Problem conclusion

• This information should be included in the Comparison of RS
  Secondary Servers and HDR Secondary Servers topic of the
  Administrator's Guide.
  
  In a high-availability cluster, the HDR secondary server must
  be the most log-advanced server among the non-shared disk
  secondary servers (RS secondary and HDR secondary). If the HDR
  secondary server becomes off line, the primary server continues
  to send logs to the RS secondary servers. However, when the HDR
  secondary comes back on line, IDS stops sending logs to RS
  secondary servers and prioritizes sending logs to the HDR
  secondary server until its log replay is ahead of the RS
  secondary server. This prioritization of the HDR secondary
  server logs is required because the HDR secondary server is the
  first failover choice in the cluster. If the RS secondary server
  logs are ahead of the HDR secondary server logs when a failover
  occurs, then the RS secondary server cannot synchronize with
  the new primary server.
  

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:

Fix information

Applicable component levels