Hello Larry, 
 
The limit of 170 buffers is equal to the 256kb configuration.  Each buffer
is 1538 bytes (buffer size we need to store an Ethernet frame), so the
1024kb configuration will provide 681 buffers.  These buffers are used for
Direct Memory Access (DMA) from the network card to Memory (as DMA applies).
The "high water" mark for the buffer pool used is "REAL
MEMORY" not virtual, of course even with the maximum configuration for the
inbound buffer pool, we are only talking about 2mb of "REAL MEMORY" if the
buffer pool reaches the maximum.
 
Normally if a ":netcontrol stop" is successful and then re-start the network
with "netcontrol start;net=loop" & "netcontrol start;net=<ni>" will be
sufficient to increase the buffer pool size, but if the circumstance for
increasing the buffer pool is leaked buffers, then a system restart will
likely be necessary.
 
Yes, you would want to run nettool;info="res;di;quit" after the problem is
seen.
 
Regards,
  James Hofmeister
 



  _____  

From: Larry Simonsen [mailto:[log in to unmask]] 
Sent: Wednesday, September 27, 2006 9:36 AM
To: [log in to unmask]
Cc: [log in to unmask]
Subject: Re: network stoppage problem



RELEASE: C.60.00   MPE/iX HP31900 C.16.01   USER VERSION: C.60.00 

from hpswinfo 
NS CORE            NSCORE     B.02.00   
NS SERVICES ASMBLY NSS        B.02.00 
NS TRANSPORT       NSXPORT    B.06.00 
NS CONFIG FILES    NWONSCF    A.02.00 
NS3000/XL NET SRV  HP36920    B.00.10 
NET SOCKETS        SOCKET     A.00.06 

no patches on this system for networking. 

WED, SEP 27, 2006, 11:29:07 PM 
 Item  Subsystem   Name   G/N  Description      Used   High    Max 
____________________________________________________________________ 
  
  1    NS XPORT  CP_POOL_ (G)  Control Buf Pool    0      1    300 :) 
  2    NS XPORT  1536____ (G)  Inbound Buf Pool   32    208    213 :) 
  3    NS XPORT  LAN1____ (N)  Outbnd  Buf Pool    2     50   2000 :) 
  4    NS XPORT  LOOP____ (N)  Outbnd  Buf Pool    0      4    128 :) 
  5    NS XPORT  UDP      (G)  GProt  Msg  Pool   50    N/A    512 :) 
  6    NS XPORT  PXP      (G)  GProt  Msg  Pool    1    N/A    660 :) 
  7    NS XPORT  PROBE    (G)  CM Prot Msg Pool    1    N/A    678 :) 
  8    NS XPORT  IP_NI    (G)  IP-NI  Msg  Pool   50    N/A   2048 :) 
  9    NS XPORT  IP_NI    (G)  IP-NI  Msg  Pool   50    N/A   2048 :) 
 10    NS XPORT           (G)  Node  Name Cache    0      1    360 :) 


yesterday i changed the input buffer size to 1024 from 320 and stopped and
started the network.  does not look like it changed the Max on the resource
display. 

when this happens on the 3000 pings are not successful.  outbound pings have
not been tried.  stopping and starting the network clears the problem. 
i will ask the remote operations people to capture this information when
this happens again ( most likely next Sunday). 



* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *