OES2 Virtual Server (e.g. clustered NSS volume) no longer visible

  • Document ID:7007479
  • Creation Date:05-Jan-2011
  • Modified Date:25-Apr-2014
    • Micro Focus Products:
      Cluster Services
      Open Enterprise Server

Environment

Novell Open Enterprise Server 2 (OES 2) Linux Support Pack 2
Novell Cluster Services 1.8.7

Situation

Intermittently, you may encounter any of the following situations:
  1. Errors during drive mapping to the clustered NSS volume during login script processing
  2. From a Windows workstation running a current Novell Client configured appropriately for Service Location, the Windows Explorer's "NetWare Servers" selection will not display the Virtual Server that is hosting the NSS volume.
Within /etc/opt/novell/eDirectory/conf/nds.conf, the"n4u.nds.advertise-life-time" parameter is set to a value lower than 1800 seconds.

Resolution

  1. Proceed to offline all mounted NSS volumes being hosted on a particular OES2 server.
    • For example at the cluster node console, where "volumes-resource" represents an NSS volume, issue cluster offline volume-resource
  2. Execute the following commands at the OES2 server console in this order:
    1. ndsconfig set n4u.nds.advertise-life-time=1800
    2. rcndsd restart
    3. rcslpd restart
  3. Proceed to online the same NSS volumes to mount them again.
    • For example, cluster online volume-resource
  4. Proceed to the other OES2 servers where this is occurring and repeat steps 1 to 3.

Note.  Instead of off-lining in step 1, you could migrate to another server, and then in step 3, you could migrate back to the original server. This option could help lessen business downtime.

Update:  A fix was released in the May 2013 OES Maintenance update (Package: novell-ncpenc ) which honors the nds.conf setting of "n4u.nds.advertise-life-time= <seconds>"

Additional Information

While the default value is n4u.nds.advertise-life-time=3600, which represent 1hr, the bindery.novell slp registration for a cluster resource (virtual server) does not renew any quicker than 30 min regardless of advertise-life-time setting. This should not be confused with a physical OES2 server that will obey the "n4u.nds.advertise-life-time" setting.

So for example, if n4u.nds.advertise-life-time is set to 600 seconds, the bindery.novell slp registrations for cluster resources will have a life-time of 600 seconds, but  they only renew every 1800 seconds.  The registration will exist for the first 10 minutes, will disappear for 20 minutes, and then come back for another 10, etc.

Therefore on a server hosting clustered volumes, an optimal setting would be to use at least n4u.nds.advertise-life-time=1800 within nds.conf.