HP3000-L Archives

June 1995, Week 1

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Eero Laurila <[log in to unmask]>
Reply To:
Eero Laurila <[log in to unmask]>
Date:
Fri, 2 Jun 1995 16:16:05 GMT
Content-Type:
text/plain
Parts/Attachments:
text/plain (39 lines)
Tom Emerson ([log in to unmask]) wrote:
: While not actually related to user limits, we recently found out that
: setting the number of active sockets too low would result in users not
: being able to log on well before you reached either your licensed limit
: or even the currently-set max session limit.  In this particular case,
: all of our users are connecting via a network using WRQ's reflection
: network series for windows.  Users would then :DSLINE out to other
: computers.  We found that both the incoming and outgoing sessions used
: a "socket", so the default (minimum) configuration of 128 sockets only
: allowed about 30 sessions to log on.
 
- No offence meant, just a little correction.  The number of "sockets"
  you are referring to is really the max allowed number of TCP-connections
  (I assume you are referring to NMMGR screen NETXPORT.GPROT.TCP).
  This value is loaded in TCP KSO (Known System Object), KSO number #333 and
  TCPSIP.NET.SYS code will check it every time an inbound or outbound
  connection is attempted to see whether it should allow or deny the new
  connection.
 
  I'm surprised to hear that in your case only 30 sessions were able to
  logon although 128 TCP connections were configured.  I can see two possible
  reasons for it:
 
    1) Since any NS server requires only one TCP connection, I would think
       that you may have some other applications (such as unispool, other
       client/server) that open TCP connections.  Maybe unispool spools to
       20 different systems and thus chews up that many TCP connections???
 
    2) There have been couple of bugs within last year where not all sockets
       were closed (and TCP connection entries released) by some NS servers.
       You may also have some 3rd party or home-grown software that has
       similar bug.  Most often those bugs have been results of people coding
       for TCP's graceful release of a connection without realizing that it
       will require two calls to IPCSHUTDOWN to completely kill the connection.
 
 
Cheers,
Eero Laurila - HP CSY Networking lab, NS services.

ATOM RSS1 RSS2