Resolving GroupWise performance issues with NSS volumes

  • 7000853
  • 07-Jul-2008
  • 27-Apr-2012

Environment

Novell GroupWise 5.5
Novell GroupWise 6
Novell GroupWise 5.5 Enhancement Pack
Novell NetWare 5
Novell NetWare 6
Novell Clustering Services
NSS

Situation

High utilization on GroupWise server
Maximum number C/S threads used on POA
POA adding C/S threads itself
POA File Queue handling extremly slow
Slow performance of GroupWise client
Clients disconnecting from POA
Resolving GroupWise performance issues with NSS volumes

Resolution

The following startup options for NSS should be considered used when running GroupWise on an NSS volume.  Note that increasing the NSS CacheBalance takes memory away from the traditional filesystem. See Performance, Tuning and Optimization for more NSS tuning recommendations.
NetWare 5.x:
/NameCacheSize=20000 /OpenFileHashShift=15 /ClosedFileCacheSize=50000 /CacheBalance=65
Note: Some options depend on the available memory in the server. If more than 1 GB of memory is available, then OpenFileHashShift and ClosedFileCacheSize can be increased to 17 and 100000 respectively.

NetWare 6:
/NameCacheSize=20000 /OpenFileHashShift=15 /ClosedFileCacheSize=50000 /CacheBalance=85 
Note: Some options depend on the available memory in the server. If more than 1 GB of memory is available, then OpenFileHashShift and ClosedFileCacheSize can be increased to 17 and 100000 respectively. Turning off allocahead on NetWare 6 is not necessary to increase performance.

(Cache Balance refers to Cache Balance Percent in Monitor | Server Parameters | Novell Storage Services)

With 6.5 and all later versions of NetWare, tuning NSS will not be necessary as the cachebalance will default to 85% and all other tuning parameters that may affect GroupWise will be dynamically adjusted by NSS.  You do not need to change the numbonds with NetWare 6.5
The /nosalvage switch should also be used on GroupWise volumes so that they do not fill up with deleted message files. You have to set this one time on the server command prompt. The syntax is /nosalvage=volumename. This has a similar effect as the PurgeImmediate attribute on traditional volumes. Once this is set, it will stick, even after rebooting the server.
The FileFlushTimer and BufferFlushTimer switches should not be used as they can cause clustered volumes to go comatose on migration and possible data loss in the event of an ABEND on the server.
The "Flush Files Immediately Attribute " for Volumes should be disabled. Enabling the " Flush Files Immediately Attribute " will decrease the performance dramatically. nss /NoFlushFilesImmediately

Additional Information

When using the defaults after installing the server and creating an NSS volume GroupWise can have big performance issues. Some fine-tuning of NSS can boost the performance. This is particularly true with NSS on NetWare 5.x.
The switches can be applied by creating a file called NSSSTART.CFG in the C:\NWSERVER folder and put the desired parameters into it as follows:

/NameCacheSize=20000
/OpenFileHashShift=15
/ClosedFileCacheSize=50000
/CacheBalance=85

Alternatively, use the server command line argument of -z"parameter...". For example:
server -z"/namecachesize=xxx /cachebalance=yy /numbonds=zzzzz"
Adding the NSSSTART.CFG file with this configuration to a NetWare 6.5 server will cause NSS volumes to fail to mount. ERROR: "zRootKey=20001"
Formerly known as TID# 10065215