We are connecting DOS based VLM clients to a NW6 sp3 server via pure IPX
and the login processes are fine. Then, NW server will give each client
a connection number(e.g from 1 to your max licensed connections which
can be seen from MONITOR).
What makes a difference (compared with NW4.11)is when we "hard reset" a
client(lose of power, warm reboot) then login again, NW6 server will
issue a different connection number to the client rather than keep the
same old number(of course for a certain MAC). Then here comes our
headache: As the files opened by the previous connection was not
immediately released by the WATCHDOG, the dead connections are still
showing in the MONITOR with opened resources, the new login won't
complete successfully when use of those resources are needed unless we
clear the previous connection or wait for WATCHDOG to disconnect them.
In NW4, when same thing happens, NW4 server will retain the same
connection number for the same MAC and upon the execution of the new
login.exe, the locked resources will be automatically released! In NW6
when scripts reached those locked files, an error message of "Network
Sharing voliation ..." happens.
Is there any other method to let server realse the dead connections
faster (except for the watchdogs) or let it handle the connecions like
NW4.11?
Please suggest, thank you.
Or you can try this util. It wasn't designed for NW6x, though, so I'm
not certain if it will work on that platform. Maybe if you turn on the
old type watchdog packets...
http://www.netwarefiles.com/utils/zzclr.zip
--
-Barry. [Novell Support Forum SysOp]
http://support.novell.com/cgi-bin/search/searchtid.cgi?/2966269.htm
I'll give both methods a try and let you know the results.
Why we what the NW server to handle connections that way is because we
noticed on the 4.11 server, it DO maintain the same connection number
for a certain MAC address and when the client executes the login.exe,
the server will release previously locked files automatically despite of
WATCHDOG settings.
Why this issue is important is because when a client is suddenly "gone",
and user tried to login again, he needs to wait a certain period of time
before WATCHDOG clears his previous login. This will generate helpdesk
support tasks when there's a large client base.
The utility is cool but I'm not to load it on server for the moment as
it will handle over 500 connections concurrently.
If so, it can't be changed back without re-introducing problems for terminal
server environments (and other cases where multiple users are logged in
using the same IP address, such as some VPN/NAT configurations).
-Sandy