Experimental Physics and
Industrial Control System

[email protected] (Jeff Hill) · Fri, 6 Nov 1998 14:02:41 -0700

William,

You should see the clients rapidly backoff to a very reasonable plateau period 
between search broadcasts if the state of the network and servers does not change.
If the state of the network and the servers does not change for 20 min then you should not see
any search requests on the network from the client that has been waiting 20 min. A change in the 
network and servers is defined as any change that causes the clients to see a new server beacon 
or a significant change in the period of an existing server beacon. 

It seems reasonable to pursue catching the special cases you identify below in the
database so that CA channels are not created. This is assuming that no one will need a  
channel named "NPP".

Nevertheless, when I run "spy" on our IOCs I have so-far not seen an IOC that was
using any significant amount of CPU in the "CA UDP" task (which would indicate that
the search requests are having an impact). I would be interested in any evidence seen
at other sites that indicates otherwise.

Jeff

On Friday, November 06, 1998 1:13 PM, William Lupton [SMTP:[email protected]] wrote:
> Dear all,
> 
>   I was just running a portable CA server with a sufficiently high debug
> level to see all the CA name resolution broadcasts (e.g. "excas -d3")
> and there are a heck of a lot. I think that the number is higher than I
> expected for two reasons:
> 
> 1. link fields of the form " NPP" etc. (which are arguably a bug but
>    which can happen if the Capfast value is something like "$(var) NPP"
>    and $(var) is blank) result in a search for a PV called "NPP"
> 
> 2. INP/OUT fields for records that are currently simulating still result
>    in name resolution broadcasts for the referenced channels (if you are
>    simulating, the referenced channels are quite likely not to exist)
> 
>   Are either of both of the behaviors expected and/or easy to change?
> 
>   Thanks,
> 
>   William
> 

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System