Experimental Physics and
Industrial Control System

"J. Lewis Muir" <[email protected]> · Fri, 26 Jul 2013 17:00:12 -0500

On 7/26/13 9:32 AM, Goff, Steven J wrote:
> I would like to use one CA application or MEDM display that will connect to two identical  servers running the same PVs. I would like both servers to receive the Channel Access ca_put at the same time. This might mean I have to modify EPICS code for each server to receive the data simultaneously.
> 
> Any ideas on how I would achieve this?

Hi, Steven.

Emmanuel Mayssat suggested a solution at the network layer.  Here's a
possible solution at the application layer: Write a custom CA server
that is a CA client to the two servers.  Your CA application or MEDM
display would connect to the CA server for the ca_put which in turn
would do the ca_put on the two identical servers.  For a ca_get, you'd
have to decide how it should work.  The CA server could treat one of the
two identical servers as a "master" and use its values, it could keep a
cached copy of the values, it could report a value status of INVALID if
the two don't agree, etc.

Of course, the term "simultaneous" is relative here.  The two identical
servers will receive the data very close in time to one another, but how
close would have to be measured, and you'd have to decide whether it was
acceptable.

Also, it would be helpful to know why you want to do this.  If the
answer is that you want some kind of redundancy scheme to improve
reliability, then this approach of having a CA server that is also a
client to the two identical servers is probably of no use since now this
CA server would be a single point of failure.

 From what I can tell, there also seems to be some people running
redundant IOCs.  Could this work for you?  You referenced it in your
Tech-Talk post in February:

  http://www.aps.anl.gov/epics/tech-talk/2013/msg00215.php

Lastly, you could give more information about what problem you're trying
to solve.  It could be that there is a better way to solve that problem
than what you've asked about.

Lewis

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System