I'm attempting to follow the instructions in the online documentation (Novell Documentation) for enabling active/active clustering of NWFTPD on a NW65SP7 cluster. The old admin had the NWFTPD tied to a single resource. Whenever that resource failed, you couldn't FTP to other volumes. I've been asked to make it so that any cluster-enabled volume can maintain it's own FTP instance.

In my test lab, I followed the instructions; Created the ftpserv.cfg files on the individual shared volumes, added the -c and -u load lines in the appropriate spots of the load/unload scripts. When a resource fails over, the ftp instance does not appear to work. I see this error message on my logger screen when the resource is coming online:

Module NWFTPD.NLM load status OK
FTP Server: UNLOAD_THIS_INSTANCE parameter set in the configuration file data03:\etc\ftp\ftpserv.cfg
Unloaded the corresponding instance
java: Class JVM Invocation by an NLM exited successfully

If I check the ftpserv.cfg file, the UNLOAD_THIS_INSTANCE is indeed set to YES when I think it should be set to NO. If I set it to NO and manually reload the nwftpd instance, it works... until the next failure/migration of the resource. It seems this bit is being set incorrectly during the service migration.

If I type everything manually on the command line, everything works as designed.

I tried adding DELAY lines in the load/unload scripts before/after the commands but it didn't help.

Anybody else seeing this same behavior?