Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Mon, 31 Mar 2014 11:17:18 -0500

Hi Ralph,

This looks very interesting, probably worthy of a HowTo in the Wiki if
you can find time.

Note though that the /etc/network/if-{up,down}.d/ directories appear to
be Debian-specific; on RHEL and Fedora systems running NetworkManager
the equivalent script would go in /etc/NetworkManager/dispatcher.d/ and
gets given 2 arguments, the interface name ("eth0", "eth1" etc.)
followed by either "up" or "down".

- Andrew

On 03/31/2014 10:15 AM, Ralph Lange wrote:
> All,
> 
> As probably many of you know, running multiple IOCs on one host has an
> annoying side effect: Clients that are using that host's IP address in
> their EPICS_CA_ADDR_LIST with EPICS_CA_AUTO_ADDR_LIST=NO will only reach
> one of the IOCs - usually the one that was started last. All clients
> have to use broadcasts to reach all IOCs.
> The same is true for CA Gateway machines that are set up in a way that
> makes multiple Gateway processes serve channels into the same network.
> 
> Here's a little helper (for Linux hosts) that I recently was playing
> around with - based on an idea by Rodrigo Bongers (CNPEM, Brazil).
> 
> If you drop the attached script into /etc/network/if-up.d *and*
> /etc/network/if-down.d, it will automatically create/delete an iptables
> rule that replaces the destination address of all incoming CA UDP
> traffic on each interface with the broadcast address of that interface.
> Simple and effective: the kernel will see all incoming name resolution
> requests as broadcasts, and delivers them to all IOCs instead of one.
> 
> Note: This will not work for clients on the same host. (Adding that
> feature makes things a lot more complicated, and I like things to be
> simple.)
> If you need connections between IOCs on one host, I would suggest adding
> the broadcast address of the loopback interface (usually
> 127.255.255.255) to each IOC's EPICS_CA_ADDR_LIST setting.
> 
> Enjoy!
> ~Ralph
> 

-- 
Advertising may be described as the science of arresting the human
intelligence long enough to get money from it. -- Stephen Leacock

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System