On 5/31/19 11:50 AM, William Layne wrote:
I ran two tests with two different PV types, ENUM and LONG. I saw the same behavior as before.
In each test I ran cainfo followed by a camonitor in two different terminals, one directed at the gateway and one at
the direct network.
Cainfo for terminals directed at gateway we're using localhost:5064. The cainfo for terminals directed at direct network were using <broadcast>:<random port that ioc was using for tcp>.
The latter setting is wrong, the port number should still be 5064 (it specifies the UDP port to be used for PV name searches). Unless you are explicitly changing the port numbers it's best to leave them off, the IOC and client will always pick the right default
port.
There is usually only one situation when you need to know the TCP port number of the server for configuration purposes, and that's when you're setting up an ssh tunnel and setting the EPICS_CA_NAME_SERVERS environment variable (you aren't doing that are you?).
Camonitor for terminals directed at the gateway saw duplicate values in camonitor. The terminals directed at the direct network did not.
Did the clients even manage to connect when configured to use the direct network as described above? What did cainfo report for the Host?
I would like to see the output from "env | grep ^EPICS" for both your two terminals, although TBH I don't see how the duplicates could have anything to do with the network configuration as such. The duplicate events have to be coming from the same CA server,
and in the case you've described that's the gateway. I don't see what could generate the duplicate monitor events unless there's a weird bug in the gateway or PCAS that could cause that.
- Andrew
I left out the other environment variables, but we have EPICS_CA_AUTO_ADDR_LIST and do not see any exceptions or warnings for duplicate IOCs. The logfile for the gateway does have one exception for an unrelated
PV (which I know is duplicated between two servers).
I can also confirm that I only see one update on the main network.
When I get in tomorrow, I will run cainfo and camonitor with main network/localhost side by side.
I agree that setting EPICS_CA_AUTO_ADDR_LIST to NO may be important, but if that's the only issue it should only result in Duplicate PV warnings from camonitor, not in duplicated heartbeat monitor values on
the connected channel.
What hostname does the "cainfo" program tell you the PV is found on, and have you been seeing any warnings about duplicate PVs found?
Is your Gateway generating any warnings about duplicate PVs? I suspect they'd go into a logfile but I forget exactly how that is configured.
Can you run camonitor connected to the main network and report whether you see the same duplicate heart-beat values? Repeat that with cainfo too...
- Andrew
On 5/29/19 4:36 PM, Mooney, Tim M. via Tech-talk wrote:
You probably want to set EPICS_CA_AUTO_ADDR_LIST to "NO". Setting EPICS_CA_ADDR list doesn't restrict attention to the named IP addresses, it just adds them to the list of addresses already found in the init process.
Forgot to CC tech talk. Oops.
Yes, we have a few interfaces. I guess it would help to give some info on that. We have it setup with client and gateway both on the same machine:
Client environment settings:
EPICS_CA_ADDR_LIST=localhost; EPICS_CA_SERVER_PORT=5064;
Gateway settings:
-cip <network broadcast>
-sip localhost
IOC settings:
EPICS_CA_ADDR_LIST=<network broadcast>; EPICS_CA_SERVER_PORT=5064;
[Client] <--> ||localhost|| <--> [Gateway] <--> ||network broadcast|| <--> [IOC]
On Wed, May 29, 2019, 4:33 PM Jemian, Pete R. via Tech-talk < [email protected]>
wrote:
Does your workstation have two active network interfaces?
Some modern Linux distributions install with a hypervisor that runs the
Linux OS as a guest to that hypervisor. The hypervisor has a management
network interface and the linux OS has a separate one. Each are on the
same physical network connection.
On 2019-05-29 3:28 PM, William Layne via Tech-talk wrote:
> Hey all,
>
> We are seeing duplicate values when first creating our subscriptions
> through the CA Gateway. Is this normal? If not, does anyone know what
> might be causing it?
>
> Note: We are on EPICS V3.14.12.8 and Gateway 2.1.0
>
> We see this in our applications and in tools such as camonitor. Here is
> an example output from camonitor, for a heartbeat PV updating every second:
>
> $ camonitor IOC1_HEARTBEAT
> IOC1_HEARTBEAT 2019-05-29 20:19:51.038659 7651
> IOC1_HEARTBEAT 2019-05-29 20:19:51.038659 7651
> IOC1_HEARTBEAT 2019-05-29 20:19:52.038654 7652
> IOC1_HEARTBEAT 2019-05-29 20:19:53.038657 7653
> ^C
>
> Thanks,
> William 'Casey' Layne
>
>
>
--
----------------------------------------------------------
Pete R. Jemian, Ph.D. <[email protected]>
Beam line Controls and Data Acquisition (BC, aka BCDA)
Advanced Photon Source, Argonne National Laboratory
Argonne, IL 60439 630 - 252 - 3189
-----------------------------------------------------------
Education is the one thing for which people
are willing to pay yet not receive.
-----------------------------------------------------------
--
Complexity comes for free, Simplicity you have to work for.
--
Complexity comes for free, Simplicity you have to work for.
|
