Experimental Physics and
Industrial Control System

"Bertrand de Boisdeffre" <[email protected]> · Wed, 20 Aug 2014 10:22:40 +0300

Dear Mark, Dear Martin,

We tried again with the following option (the rest are disabled) : 
EPICS_CA_AUTO_ADDR_LIST=YES. 

And it works.

Thanks a lot.
Best, 
------------------------------
Bertrand DE BOISDEFFRE
Engineer
ELI-NP, Horia Hulubei National Institute for Physics and Nuclear Engineering
30 Reactorului Street,
077125 Magurele, Romania
+4021 404 50 89 / 021 404 50 89
www.eli-np.ro

-----Original Message-----
From: Mark Rivers [mailto:[email protected]] 
Sent: Tuesday, August 19, 2014 5:49 PM
To: 'Bertrand de Boisdeffre'; 'Martin L. Smith'; [email protected]
Subject: RE: New users - Channel Access Clients issues

If the IOC and medm are running on the same machine then I would suggest not
setting the EPICS_CA environment variables at all.

At the IOC prompt you can see what the environment variables are with this
command:

epicsPrtEnvParams

Make sure iptables is stopped and restart both the IOC and medm.

You can also do a simple test with the "caget" and "cainfo" programs from
EPICS base.  For example:

/usr/local/epics/base-3.14.12.4/bin/linux-x86_64/caget 13IDA:m1
13IDA:m1                       0.0448

>/usr/local/epics/base-3.14.12.4/bin/linux-x86_64/cainfo 13IDA:m1
13IDA:m1
    State:            connected
    Host:             ioc13ida.cars.aps.anl.gov:5064
    Access:           read, write
    Native data type: DBF_DOUBLE
    Request type:     DBR_DOUBLE
    Element count:    1

Mark

-----Original Message-----
From: Bertrand de Boisdeffre [mailto:[email protected]]
Sent: Tuesday, August 19, 2014 9:34 AM
To: Mark Rivers; 'Martin L. Smith'; [email protected]
Subject: RE: New users - Channel Access Clients issues

Hi Mark, Hi Martin,

Actually we are running the IOC and the Client on the same computer (we are
beginners in EPICS and trying to understand how it works).
Before Martin sent me the message, we thought already about disabling the
EPICS_CA_ADDR_LIST option and enabling the other,
EPICS_CA_AUTO_ADDR_LIST=YES.

Apparently (one of my colleague made this modification), this didn't solve
our problem. 
We will try to redo it again tomorrow and I will send you the feedback.

Thanks a lot for your help,

Best,

------------------------------
Bertrand DE BOISDEFFRE
Engineer
ELI-NP, Horia Hulubei National Institute for Physics and Nuclear Engineering
30 Reactorului Street,
077125 Magurele, Romania
+4021 404 50 89 / 021 404 50 89
www.eli-np.ro

-----Original Message-----
From: [email protected] [mailto:[email protected]]
On Behalf Of Mark Rivers
Sent: Tuesday, August 19, 2014 5:10 PM
To: 'Martin L. Smith'; [email protected]
Subject: RE: New users - Channel Access Clients issues

Hi Bertrand,

What is the network configuration of the IOC machine and the medm machine?
Are they running on the 

1) Same computer
2) Different computer on same subnet
3) Different computers on different subnets?

If 1) or 2) then you should not set the EPICS_CA_ADDR_LIST environment
variables at all, the defaults should work.

If 2) then you may need to disable iptables on both machines.

> Stopping iptables did manage to solve some port issues (when starting 
> the ioc the shell gave a warning about the default port being 
> unsuitable
and giving a random port) but the problem with the ./caExample persists and
MEDM is still not getting any values.

The warning you saw is normal if you have 2 IOCs running on the same
machine.  Disabling iptables should have no effect on that warning. I
suspect the warning went away for a different reason, i.e. that you were no
longer running 2 or more IOCs.

Mark

-----Original Message-----
From: [email protected] [mailto:[email protected]]
On Behalf Of Martin L. Smith
Sent: Tuesday, August 19, 2014 8:27 AM
To: [email protected]
Subject: Re: New users - Channel Access Clients issues

Hi Bertrand,

I assume you are using the EPICS_CA_ADDR_LIST and EPICS_CA_AUTO_ADDR_LIST
because your CA client is not running on the same network as the IOC.

I think that if you do not have a firewall between your networks but there
is in fact a network route through a router then the above environment
variables should take care of the CA client side connections but the IOC
must know how to send information back to the client.

For that you will likely need to have similar environment variables set in
the IOC st.cmd file. Otherwise you could run a PV gateway between the
networks.

Also if you have EPICS_CA_AUTO_ADDR_LIST=YES then you likely do not need to
have the broadcast address 192.168.163.255 in your EPICS_CA_ADDR_LIST;
however, you might need to specifically include the IP address of the router
for both the CA client as well as the IOC.

Marty

On 08/19/2014 08:07 AM, Bertrand de Boisdeffre wrote:
> Dear EPICS users,
>
> We installed Epics base 3.14.12.3 on a computer using Scientific Linux.
>
> Then we managed to create, build and run the IOC example application 
> included with the EPICS base.
>
> Now, we would like to create one CAClient to access the PV.
>
> Then we configured the /etc/profile file and added the necessary lines:
>
> EPICS_HOST_ARCH="linux-x86_64"
>
> export EPICS_HOST_ARCH
>
> EPICS_BASE=/opt/epics/base-3.14.12.3
>
> export EPICS_BASE
>
> PATH=$PATH:/opt/epics/base-3.14.12.3/bin/linux-x86_64
>
> EPICS_CA_AUTO_ADDR_LIST=YES (tried this too, but it did not work)
>
> export EPICS_CA_AUTO_ADDR_LIST
>
> EPICS_CA_ADDR_LIST="192.168.163.97 192.168.163.255"
>
> export EPICS_CA_ADDR_LIST
>
> and we am unable to reach any PV in the example app included in Epics.
>
> We am running the example app by running st.cmd in terminal, we can 
> see the PVs inside Iocshell but when I try to connect the PV to a 
> graphical text monitor in MEDM, the Graphical Object is not connected 
> to
the PV inside the IOC.
>
> PV Info inside MEDM says:
>
> TYPE: TYPENOTCONN
>
> HOST: <disconnected>
>
> The PV name is correct, we think: rootHost:ai1.VAL and tried 
> rootHost:aiExample1.VAL and they are not receiving any values from the
IOC.
>
> When we try to run caExample we get the following errors:
>
> [root@Epics O.linux-x86_64]# ./caExample rootHost:ai1.VAL
>
> CA.Client.Exception...............................................
>
>      Warning: "User specified timeout on IO operation expired"
>
>      Context: "ca_pend_io failure"
>
>      Source File: ../caExample.c line 20
>
>      Current Time: Tue Aug 19 2014 14:54:46.378283619
>
> ..................................................................
>
> CA.Client.Exception...............................................
>
>      Warning: "Virtual circuit disconnect"
>
>      Context: "ca_get failure"
>
>      Source File: ../caExample.c line 21
>
>      Current Time: Tue Aug 19 2014 14:54:46.378688325
>
> ..................................................................
>
> CA.Client.Exception...............................................
>
>      Warning: "User specified timeout on IO operation expired"
>
>      Context: "ca_pend_io failure"
>
>      Source File: ../caExample.c line 22
>
>      Current Time: Tue Aug 19 2014 14:54:51.378812388
>
> ..................................................................
>
> rootHost:ai1.VAL 0.000000
>
> We thank you in advance for your support.
>
> Best regards,
>
> ------------------------------
>
> Bertrand DE BOISDEFFRE & Vlad RUSU
>
> ELI-NP, Horia Hulubei National Institute for Physics and Nuclear 
> Engineering
>
> 30 Reactorului Street,
>
> 077125 Magurele, Romania
>
> +4021 404 50 89 / 021 404 50 89
>
> www.eli-np.ro <http://www.eli-np.ro/>
>

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System