Environment
Novell NetWare 6.0 Support Pack 5
Novell NetWare 6.0 Support Pack 4
Novell NetWare 6.5 Support Pack 2
Novell NetWare 6.5 Support Pack 3
Novell NetWare 6.5 Support Pack 4
Novell NetWare 6.5 Support Pack 5
Situation
The cluster join will sometimes hang the system
console. The server will experience high utilization
during this period. A reboot of the node may or may not
successfully allow the node to join. Disk IOs appear to be
stuck.
The console command process is stuck in a tight loop waiting for the CDMActivateCount to go to null. When attempting to join a cluster, we will first do a "scan for new devices" in an effort to find the SBD or Cluster partition. If there are a lot of LUNs the CDM_Activate_Count may get off causing the hang.
The console command process is stuck in a tight loop waiting for the CDMActivateCount to go to null. When attempting to join a cluster, we will first do a "scan for new devices" in an effort to find the SBD or Cluster partition. If there are a lot of LUNs the CDM_Activate_Count may get off causing the hang.
Resolution
The customer can decrease the likelihood of this issue occurring by
decreasing the number of IO requests that are made during a "scan
for new devices". In the case of a GLogic driver, you
can use the "maxluns" setting when loading the driver to
decrease the number of LUNs that must be scanned we also removed
the "inquiry" parameter in an effort to decrease the number of IO
requests being made to the SAN. The long range solution is to
upgrade the cluster to NetWare 6.5 SP6 or higher as the bug has
been fixed in NWPA.NLM version 3.21 dated September 26, 2006 which
was included in NetWare 6.5 SP6.