Experimental Physics and
Industrial Control System

Markus Kristensson via Tech-talk <tech-talk at aps.anl.gov> · Thu, 27 May 2021 14:44:09 +0200

Hej Torsten,
Hope you are doing well!

Interesting. Since I use PyEpics/PyQt for the GUI that would probably be

a solution for me. However, I do use e3, but I suspect that updating

this make-file would be feasible herein also?

For future reference, if pvaccess was to be used on an offgrid device,

could a possible solution be to create a virtual dummy interface on the

RPi?

Cheers from Skåne,
Markus

2021-05-27 08:41 skrev Torsten Bögershausen:

Hej Markus,

as far as I know:
The "IOC not probably working when we don't have any physical network
interfaces" is only there for pvacess.
pvaccess uses a different way than channel access
(to detect interfaces which the ca/pv binds to).
Looking from the EPICS community point of view, it seems as if this
"networkless IOC" is rarely used.

In your case it may help to disable the pvaccess completely.

I have something like this in my Makefile:
ifdef BASE_7_0
 ethercatmc_DBD += PVAServerRegister.dbd
 ethercatmc_DBD += PVACommonRegister.dbd
 ethercatmc_DBD += softIocPVA.dbd
 ethercatmc_DBD += registerChannelProviderLocal.dbd
endif

HTH, greetings to Lund
/Torsten

On 5/27/21 12:04 AM, Markus Kristensson via Tech-talk wrote:

Hi again,

I've investigated the problem further. It turns out that I was too

quick in judging that the IOC was starting up more or less correctly

even when the ethernet cable was not plugged in. The reason for that

conclusion was two-fold: 1. I ran the RPi in headless mode without any

monitor connected. 2. The RPi did respond to change in some hardware,

namely me changing a physical switch between two states. The response

of changing the switch was seen in the opening/closening of a relay.

As already stated, the response was very slow (sometimes around 10 s).

What bugged me was that the same relay card did not open when a number

of other DI:s received signal.

With a monitor connected I now realise that the IOC is actually NOT

starting correctly when the ethernet cable is unplugged. The error is

a BaseException with "No (specified) network interface(s) available".

Interestingly, the plc (I run the ecmc-module) which handels the logic

starts up briefly before the IOC startup is terminated. This gives

enough time to read the input from the switch and to open up the

relay, but not enough time for the other inputs to respond because of

the logic using hysteresis (this is a solution for dealing with 14

cooling rooms and 2 cooling machines etc...). The slow responding time

of the switch (max around 10 s) should then basically be related to

the restart time of the IOC which I've configured using systemd to 10

s.

So, some things better understood, but my basic problem still remains.

Do you have any suggestions of how I can resolve the problem of

starting an IOC without it being connected to any network? I suspect I

need to modify the startup script of the IOC, e.g. specifying

EPICS_CA_ADDR_LIST to localhost. So far I have not been successful

with such modifications.

Cheers,
Markus

2021-05-26 20:31 skrev Markus Kristensson via Tech-talk:

Hi,

I have configured a Raspberry Pi 3 Model b+ to autostart and launch

an

IOC. The startup works fine if the RasPi is connected via ethernet to
another computer. If the lan is not connected at startup, the IOC is
basically unusable, responding to certain inputs, but extremely slow.
If the ethernet cable is plugged back in, the responsiveness of the
IOC is back to normal.

The RasPi will normally be connected to the other computer, but it is
not necessarily so. I only need to control and access the IOC on the
RasPi locally (i.e. on the Raspberry Pi itself). Do you have any
suggestions how to proceed?

Cheers,
Markus Kristensson

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System