Conference netcad::hub_mgnt

    Is the customer intending to operate the 10BaseT Ethernet ports in
    half or full duplex mode?
    
    Hugh
    

    re .-1 
    	They use half duplex
    
    	VNSwitch revisions:
    
    	HW 06f1e1, RO V1.5.4, SW V 1.5

    How many of the Ethernet ports are connected?  Is the rate of
    transmit errors related to the number of active Ethernet ports?
    
    Hugh
    

Hi,
Since the VNSwitch was replaced by DECSwitches 900EF, customer is not experiment
transmit errors  anymore and everithing works fine.

What about Echo and F/W version of the VNswitch as mentionned above ?

Thanks,
Robert

    A common problem that generates transmit errors is when one side of
    the link is running full duplex and the other is running half duplex.
    If both sides are running half duplex, that is not the problem.
    
    The version numbers give no hint of the problem.
    
    In .3, I was just trying to understand if the error rate was traffic
    load related.
    
    Hugh
    

Hugh,

The customer is coming from an environement running FDDI and Ethernet,
most of the comment equipments were DECConcentrator 500 and DECBridge 500 also
several DEHUB900
are used to connect Ethernet to FDDI (equiped with DECSwitches 900EF)
This is an important site where we have sold a GigaSwitch FDDI and VNSwitch 
in order to replace the old DEConcentrators and bridge and to anticipate
network load.
Since that time the customer was claiming about slow response time and NFS
time-out, especially for the
users that are connected to the VNSwitch900EF. So the customer plug some of
them back again to the DECbridge500
and the problem was gone.
As a interim solution we proposed to replace the VNSwitch900EF by 2
DECswitch900EF which are currently installed
and everything seems to working fine (I am going to day at the customer site
to verify this).

So there is no lot of flexibility to play at this customer site and I would
like to have a statement about
performance issues on VNSitch and what to tell to the customer.

If I am allowed and have some time I will upgrade it with version 1.54.
My main concern is ECO version or H/W problems and if this is the issue I
don't want to keep 
me busy (and the customer aswell) with useless firwmare update and cabeling
reconfiguration.


Thanks in advance,
Robert Vandenberghe


[Posted by WWW Notes gateway]

    Robert,
    
    There isn't enough information for me to explain the customer's
    complaints.
    
    We have fixed a number of software problems since V1.5.0.  For some
    customers, crashes were occurring without the customer being
    aware of the system restart -- the network would just become
    unavailable for about 2 or 3 minutes.  So you might want to check
    the error logger.  If you see some crashes, definitely upgrade.
    
    During INTENSE translation load tests (Ethernet<->FDDI) we have
    seen transmit errors associated with transmit underruns.  This
    may happen occasionally in operational networks, but we do not
    expect it to happen at a significant rate.
    
    I am not aware of any other time out/performance issue assocaited
    with the VNswitch EF.  We are capable of translating about 180K
    packets/second -- about 3 times more than the DECswitchEF.
    
    Hugh

FYI,
Firmware was upgraded to 1.5-004
The customer will progressively reconnect the systems to it, nevertheless, we
can already monitor lots of Recieve Discard Frames at the FDDI interface. This
is not the case for DECSwitch900 EF that are connected on the same ring.
Any idea about this ?

Robert Vandenberghe

    Can you indicate exactly which counters are intrementing?
    
    There may be different behavior between the two switches or
    different thing may be counted.
    
    Sorry I can't be more helpful.  There isn't enough information.
    
    Hugh