Comments on OPC UA : WCCOAopcua process weird behaviour

[a.decelis]

Hi,

I've been testing for some time, connections to (third party) OPC UA Servers with two different versions of WinCC OA: 3.12 (P020) for Windows (W7_64) and 3.13 (unpatched) for Linux (CentOS 7 x64). And I have found some funny behaviours I would like to comment.

First point: if I test against an insecure OPC UA Server, that is, one not using any certificates, the manager (WCCOAopcua) aborts abruptly. In both Windows and Linux.
Under Windows, a Windows error message informs of the occurrence of exception c0000005, that is, (Memory) Access Violation.
I am not saying, neither it's a good idea to use unsecured servers (is it perhaps a violation of the binary procotol?) nor WCCOA should allow it. But, I can't find an explanation to why do WCCOAopcua processes die upon connecting to an insecure server. Not sure if a bug or a feature, just did not find any documentation regarding this behaviour.
Wouldn't it be more reasonable if the process just disconnected and emmited an error message?

Second point, scenario: OA 3.12 under Windows, after having succesfully configured a secure connection to an OPC UA server.
Upon restarting WinCC OA project, the WCCOAopcua process fails several times with a c0000005 exception. Eventually, after two failed respawns, the third one succeeds and the process survives, and connection to the UA server goes smooth after that.

Third point: There is a note in the documentation which states that Simulation Driver cannot run concurrently with the OPC UA client.
Well, practice asserts that to be very true: running the simulation driver prevents the opcua client to start. Ok, right.
I would have found much more useful an error message stating that, rather than the actual message received (Connection error, see [1]).
I understand that the simulation driver has taken then role of the UA client, thus preventing it to start, but the message does not (directly) suggest that. And it took me some time of head scratching and reinstalling, until a fellow pointed it out for me.

Thanks for your attention.



[1]
WCCOAopcua (1), 2015.08.19 13:55:30.044, SYS, INFO, 3, Trying to connect to, (SYS: 0 Data -num 0 CONN: 1) @ localhost:4897
WCCOAopcua (1), 2015.08.19 13:55:30.047, SYS, INFO, 4, Connected to, (SYS: 0 Data -num 0 CONN: 1) @ admin-THINK (::1)
WCCOAopcua (1), 2015.08.19 13:55:30.048, SYS, INFO, 39, Connection lost, MAN: (SYS: 0 Data -num 0 CONN: 1), Connection closed
WCCOAopcua (1), 2015.08.19 13:55:30.049, SYS, FATAL, 63, Connection error, This manager is already connected or Data is running in emergency / backup / recovery mode or the active data is unreachable.
WCCOAopcua (1), 2015.08.19 13:55:32.153, SYS, INFO, 2, Manager Stop

[leoknipp]

Hello,

if the problem "First point" and "Second point" still occur when using the newest summary patches for WinCC OA please get in contact with your common WinCC OA support and report the problem.

The "Third point" is not a problem it is the correct behaviour.
WinCC OA processes of a specific manager type can only be started once with the same manager number. The simulation driver and the OPC UA client are both of manager type "drivers" and therefore they have to be started with a unique number if they shall run at the same time. E.g. simulation driver with number 1 and OPC UA client with number 2.

Best Regards
Leopold Knipp
Senior Support Specialist