Experimental Physics and
Industrial Control System

"Manoussakis, Adamandios via Tech-talk" <tech-talk at aps.anl.gov> · Fri, 19 Feb 2021 00:11:56 +0000

Thanks for everyones input, will keep debugging I am guessing it must be some sort of routing issue or not setting an env variable correctly.  

Does anyone have a good example of when you would use the CAGateway and CARepeater?  
Also does doing broadcast addressing like 192.168.1.255 the same as implementing the CARepeater?

-----Original Message-----
From: Michael Davidsaver <mdavidsaver at gmail.com> 
Sent: Thursday, February 18, 2021 2:03 PM
To: Manoussakis, Adamandios <manoussakis1 at llnl.gov>
Cc: tech-talk at aps.anl.gov
Subject: Re: Identical Process Variable names on Multiple Servers

On 2/18/21 1:06 PM, Manoussakis, Adamandios via Tech-talk wrote:
> Sorry I think I may have left out part the IPs are for example 
> 192.168.1.40 and 192.168.2.50.  They exist on different subnets and I didn’t have trouble at first running one IOC on say 192.168.1.40 and having 192.168.2.50 be the client doing caget/camonitor.
> 
>  
> 
> I have the client(192.168.2.50) env variable CA_ADDR_LIST=192.168.1.40 
> and it was working fine but now that I have switched things up and I have 192.168.1.40(CA_ADDR_LIST=192.168.2.50) be the client and 192.168.2.50 have the single IOC running it doesn’t seem to want to communicate.
> 
>  
> 
> Which led to me believe I might have just gotten lucky before when it was working and I am missing a piece of the puzzle such as using cagateway or like you have mentioned broadcast addressing.  When you stated to set both to the same broadcast address, do you mean that 192.168.1.40 machine would have 192.168.2.255 and the 192.168.2.50 machine would have 192.168.1.255 set under CA ADDR_LIST respectively.

I didn't read closely enough and was assuming that .1 and .2 were referring to two hosts on the same subnet.  My comment about broadcast addresses would only apply if both where on the same subnet.

Actually, you might check to see if your router is configured to route IPv4 multicasts between these two subnets.  If it is, then you could use the new-ish mcast search feature.  If it is, then you can add a (hopefully unused) multicast address to EPICS_CA_ADDR_LIST and also set EPICS_CA_MCAST_TTL=2

> *From:* Tech-talk <tech-talk-bounces at aps.anl.gov> *On Behalf Of *Ralph 
> Lange via Tech-talk
> *Sent:* Thursday, February 18, 2021 8:26 AM
> *To:* EPICS Tech Talk <tech-talk at aps.anl.gov>
> *Subject:* Re: Identical Process Variable names on Multiple Servers
> 
>  
> 
> On Thu, 18 Feb 2021 at 16:14, Michael Davidsaver <mdavidsaver at gmail.com <mailto:mdavidsaver at gmail.com>> wrote:
> 
>     On 2/18/21 7:01 AM, Ralph Lange via Tech-talk wrote:
>     > When using an explicit IP for the name search, only the /last/ IOC started on that machine will answer.
> 
>     FYI. first vs last depends on host OS, and as I recall windows and linux differ.
> 
>  
> 
> Hm! I was not aware of that at all. Will verify and update the README.
> 
>  
> 
>     2. mis-configured broadcast address
> 
>     Both the .1 and .2 host must have the same broadcast address set.
> 
>  
> 
> As .1 and .2 are different networks, it actually depends on the structure.
> 
>  
> 
> Always: The broadcast addresses must match between the IOC host (receiving the broadcast) network configuration and the client (sending the broadcast) ADDR_LIST setting.
> 
> If the client host has a network interface into the IOC host network: The client host network configuration on that interface must also match.
> 
> If there is a switch between the two networks: The switch needs to be configured to forward "directed broadcasts".
> 
>  
> 
> Cheers,
> ~Ralph
> 

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System