Experimental Physics and
Industrial Control System

Benjamin Franksen <[email protected]> · Mon, 31 Oct 2016 12:09:04 +0100

Hi everyone,

we finally found out what caused this strange behaviour. There was
indeed a misconfiguration in the network: a new device was mistakenly
copnfigured to use the same IP address as the IOC. For reasons that
remain mysterious to me this condition was/is not detected by our
network administrators. Instead we found out by trial and error and a
bit of luck: someone had the idea to unplug the IOC and then issue a
ping (which was answered by the 'imposter'). If the new device had been
configured to not answer pings we would still be wondering...

Cheers
Ben

On 05.08.2016 17:23, Benjamin Franksen wrote:
> """
> We have a PC with Windows 7 that has Labview and an EPICS base 3.14.12.2
> installed. This PC connects to several channel access servers. The
> servers that are Linux based, like a channel access gateway and a
> softioc, make no problems. However, since last week, connections to a
> vxWorks IOC are no longer possible.
> 
> "cainfo" for a PV on this IOC replies something like "connection reset
> by host", although the correct host is mentioned. So it seems that the
> name resolution request for the PV works but the connection to actually
> transmit the data is interrupted or can't be established at all. Both
> systems are within the same network collision domain.
> 
> A Linux channel access client in the same network has no problems with
> it's connections to the same PVs, so it doesn't seem to be a server
> related problem.
> 
> Now, if we change the IP address on the windows PC to another free
> address in the network, channel access first works. Everything is fine
> but only for 3 to 5 minutes. Then connections to the vxWorks IOC fail
> again. Setting the IP address to a third variant has the same effect.
> Everything works but only for a few minutes. Setting the IP address back
> to the original one still doesn't work. Deactivating and  re-activating
> the network interface doesn't help. After a while, maybe  15 to 30
> minutes, an IP address that didn't work and wasn't used for that time,
> can be used again, but still, only for 2 to 3 minutes.
> 
> Setting EPICS_CA_ADDR_LIST to the IP of the IOC and
> EPICS_CA_AUTO_ADDR_LIST to "no" doesn't help. But channel access on the
> windows PC is not completely broken. Requesting PVs from our linux
> based channel access gateway or a linux soft-IOC always works.
> """
-- 
"Make it so they have to reboot after every typo." ― Scott Adams

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System