Conference irocz::terminal_servers

3428.0. "DNAS v2.0, BL10C-40 100% CPU used" by CSC32::D_MATTHIES () Thu Jan 30 1997 15:21

    Hi All,
    	I have a Customer with a DECserver 900TM running v2.0, BL10C-40
    (got the patch before the hold) that shows %CPUused  of 2/100/100 and
    the Users complain of intermittent slow response.  This is a lightly
    used DECserver on a relatively quiet network based on Status and
    Counter displays.  The first 16 ports are telnet Users on terminals, no
    PC's, the next 5 port have simple printers, LA75's and the like, and
    the last 11 ports are unused.  
    	Active Ports show 12/17/32, Reachable Nodes 2/3/200, Active
    Circuits (LAT) 0/1/32, Connected Sessions 12/17/64, %MEMused 41/41/100
    and both Selftest and Software Status are Normal.  There are no Resource
    Errors and only 1 Port error on any type which was a Framming error.
    	The whole right side of the counter display is 0 with the following
    exceptions:  3 Single Collisions, 4 2+ Collisions and 29 Messages
    Re-Transmitted out of 187,087 Messages Transmitted.
    	I thought maybe we had a Broadcast Storm which caused the CPU to be
    so high but there are two DECserver 90TL's, running v1.1a, BL45-11,
    essentially right next to the DECserver 900TM and they show about the
    same Network Activity with a %CPUused of only 43% high.  All the
    DECservers have been up for about the same length of time, 4+ days, as
    the Customer rebooted them all over last weekend.
    	Does anyone know of anything else that might cause the CPU usage to
    jump up like that?  Is it possible this display is not reporting
    accurate information under this software and/or version and the slow
    reponse is caused by something else?  Any thought, ideas, suggestions,
    solutions will be greatly appreciated.
    
    David Matthies
    Open System Networking
    Customer Support Center
    
    PS:	I do not believe it makes a difference but the OS is Digital Unix
    v3.2d-1.