Experimental Physics and
Industrial Control System

"Jeff Hill" <[email protected]> · Fri, 3 Nov 2006 09:37:54 -0700

Emmanuel,

> I run an IOC on computer with 2 network interfaces.
> When I try to caget a process variable I get an error telling me that
> the PV is accessible from 2 addresses (multiple servers warning).

I could imagine seeing that result if you have configured your network
interfaces to (both of them) be on the same subnet, and caget is being run
on another host on this subnet? 

Alternatively, if you are running caget on the same multi-homed host as the
IOC then it might be necessary to explicitly configure the client's address
list so that it does not broadcast on both subnets (and find the same IOC on
both subnets)? And of course, you could just avoid running clients on
machines that are also routers. In some situations maybe dedicated routers
are cheap enough that we can avoid using workstations for this purpose?

> I tried to
> EPICS_CA_AUTO_ADDR_LIST=NO
> EPICS_CA_ADDR_LIST=<myHostname>:5064
> 
> But then I get the message:
> Empty PV search address list.
> and cannot find the PV at all.

This (the empty search address list message) is not my experience here
(testing on Linux). Admittedly I am not running on a multi-homed host, but I
don't expect that would cause the address list to be empty when it is
explicitly configured as follows.

~$ export EPICS_CA_ADDR_LIST=wombat:5064
~$ export EPICS_CA_AUTO_ADDR_LIST=YES
~$ caget joh:fred
joh:fred                       0.0066983

Jeff

> -----Original Message-----
> From: Emmanuel Mayssat [mailto:[email protected]]
> Sent: Thursday, November 02, 2006 7:53 PM
> To: epics
> Cc: [email protected]
> Subject: Multihomed IOC
> 
> Hello,
> I run an IOC on computer with 2 network interfaces.
> When I try to caget a process variable I get an error telling me that
> the PV is accessible from 2 addresses (multiple servers warning).
> 
> I tried to
> EPICS_CA_AUTO_ADDR_LIST=NO
> EPICS_CA_ADDR_LIST=<myHostname>:5064
> 
> But then I get the message:
> Empty PV search address list.
> and cannot find the PV at all.
> 
> It doesn't make any sense!
> Please help,
> --
> Emmanuel

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System