Experimental Physics and
Industrial Control System

Jeff Hill <[email protected]> · Wed, 08 Jan 1997 07:47:37 -0700

[email protected] wrote:

> >Note that the efficiency of the name resolution activity was improved
> >in 3.12 at some point (by changing a time constant). Perhaps Marty's numbers
> >do not agree with what CEBAF has observed because he is using a more recent
> >version of the code.
> 
> We are running 3.12.2.

Have you verified connect performance problems for:
o after upgrading to 3.12.2
o 2000 channels
o CA server at default priority levels
o CAMAC in normal state
o 80% loaded mv167

We have not seen a problem here when the first 4 conditions are true.
I need to experiment with condition 5.

> 
> >Note that the CA client lib _WILL_ recover in the rare circumstances when
> >this occurs if the operator is willing to wait for the full duration of
> >the timeout.
> 
> We are seeing hung behavior persisting for a MUCH longer time than 80 seconds,
> probably indicating that we encounter this timeout multiple times for a
> single MEDM screen. (Question: how many names are sent in 1 packet?)
> 

No doubt that it will try to reconnect (and therefore time out multiple 
times). Also, the default (or configured) TCP connect sequence timeout on 
the hp systems may be different.

> I think the best commendation I've heard is to raise the CA TCP task up in
> priority.  Can anyone think of any down side to this, other than compromising
> the real time nature (mythical) of the lower priority scan tasks?  If this
> is safe, I can live with the synchronous connect problem.
> 

This has not been tried. I dont know of any reason why CA would fail
if you did this. Your IOCs may be less immune to increasing network 
load in this situation (compounded when CPU loading is high). Some of your 
application specific database logic may behave differently when interrupted 
by ca during a scan. Also, this will make your systems different from other 
installations and therefore perhaps it will be more difficult to diagnose 
any problems that may occur at your site?

I am looking at a non-blocking connect implementation. A patch on
a branch off of 3.12.2 is possible if we go ahead with the change.
Would you (or anyone else) use this patch?

Jeff

-- 
______________________________________________________________________
Jeffrey O. Hill                 Internet        [email protected]
LANL MS H820                    Voice           505 665 1831
Los Alamos, NM 87545 USA        FAX             505 665 5107

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System