NLDAP can become unresponsive after dibcloning a server

  • 3203523
  • 10-Aug-2006
  • 26-Apr-2012

Environment


Novell eDirectory 8.7.3.8 for Solaris
Novell eDirectory 8.7.3.8 for Linux

Situation

Scenerio:
Large tree - greater than 4 million objects
Large amount of changes via NLDAP in the tree
Current change cache for many partitions contains a large number of objects - approx. 50,000-100,000
5 servers in the tree - 3 in an LDAP pool and 2 outside the pool
One of the servers outside the pool is cloned
As soon as servers in the rings receive an update to the partition adding the new server, partitions get flagged with"pf_rebuild_change_cache" causing a rebuild of the change cache for all partitions included on the cloned server.
Disk channel is 100% during the change cache rebuild
NLDAP is unresponsive until change cache is rebuilt for all partitions included on the cloned server.

Resolution

Suggestion:

Clone servers with large busy partitions only during times when modifications are low so that the number of objects in change cache are low.

This has been reported to development.

Status

Reported to Engineering