[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference noted::decnis

Title:	DEC Network Integration Server (DECNIS)
Notice:	Please read note 1 to use this conference effectively
Moderator:	MARVIN::WELCH

Created:	Wed Sep 18 1991
Last Modified:	Thu Jun 05 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	3660
Total number of notes:	15082

3659.0. "Any limitation on the number of CTF trace on DECnis?" by HTSC19::KENNETH () Wed Jun 04 1997 00:05

Hi,

One of my customer use CTF to monitor the X25 lines on their DECnis 600.
Everytime he only set up two traces. One live and one record to a file.

That is :

Trace start/live decnis""::"x25l3 dte det-3-0"/data=ascii/wid=132

Trace start decnis""::""x25l3 circuit fdp-prim""/process=tra$fp0/out=fp0.dat
/ver=100/capture=519"

One day the Decnis hung and after investigation by Engineering, it was 
caused by the trace and pool memory being eaten up.  Therefore Engineering
suggest the customer limit the trace buffers.

Now the customer would like to know is there any limitation on the number of
traces that we can start on the DECnis?  What is the recommand buffer size?

Is there anyone knows?

Thanks for your help in advance.

Kenneth Leung

T.R	Title	User	Personal Name	Date	Lines
3659.1	There is no simple answer	MARVIN::RIGBY	No such thing as an alpha beta	`Wed Jun 04 1997 05:51`	28
	>Now the customer would like to know is there any limitation on the number of >traces that we can start on the DECnis? What is the recommand buffer size? Unfortunately the question "what is the CTF session limit" cannot have a simple answer. There are obvious constraints 1) the amount of memory reserved for holding CTF trace records 2) the CPU overhead of the trace operation 3) the NSP or TCP connections and bandwidth consumed to maintain the trace The memory constraint is most restrictive on an MPC-I as it has half the memory available of an MPC-II. The amount of memory available on any system that can safely be consumed by trace buffers depends on how much is left after all the other memory users have had their fill. Some of these other memory uses are configurable (routing database sizing for example), some might be removed by disabling features that are not actually needed. Use the NCL command SHOW decnis DEVICE UNIT mp* NUMBER OF FREE BUFFERS to determine how much pool is available before starting the trace operations. The trace will consume max_buffers x buffer_size + some overhead. I would recommend checking if there are routing database sizing parameters that are grossly oversized for the customers needs to free up some pool memory. I would also point out that DTF (the DECNIS specific trace facility) can be used to simultaneously capture traffic to a file AND display the output to the screen while using only a single connection and set of trace buffers. John
3659.2	Thanks but one more question	HTSC19::KENNETH		`Wed Jun 04 1997 06:40`	9
	Hi John, Thanks. Do you mean that each trace consume max_buffers x buffer_size + some overhead. So I have 2 traces then I need 2 x (max_buffers x buffer_size = some overhead), is that right? Thanks again for your help. Kenneth Leung
3659.3	yes, each trace	MARVIN::RIGBY	No such thing as an alpha beta	`Thu Jun 05 1997 09:08`	5
	>Do you mean that each trace consume max_buffers x buffer_size > + some overhead. So I have 2 traces then I need 2 x (max_buffers x > buffer_size = some overhead), is that right? Yes.