Loss of DeviceNet connection

[ulrich06]

Hello,

We have successfully deployed a KUNBUS DeviceNet gateway on a production line to retrieve information from a PLC.
The gateway is detected by the DeviceNet master and we are able to retrieve data using piTest:

pi@RevPi100257:~ $ piTest -d
Found 2 devices:

Address: 0 module type: 95 (0x5f) RevPi Core V1.2
Module is present
input offset: 0 length: 6
output offset: 6 length: 5

Address: 32 module type: 73 (0x49) Gateway DeviceNet V0.0
Module is present
input offset: 11 length: 512
output offset: 523 length: 512

pi@RevPi100257:~ $ piTest -r 11,25
86 41 85 148 239 1 22 0 0 0
0 0 87 84 87 4 170 1 170 106
82 1 81 0 2

However, since we have updated the RevPi Core module from Jessie to Stretch, we regularly experience connection losses, i.e., we only read "0" values :
pi@RevPi100257:~ $ piTest -r 11,25
0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0
0 0 0 0 0

We have noticed that during these disconnections, both Power and MS LEDs turn flashing red. We can only recover the connection by rebooting the Core module.

Any hint to solve this issue? Are there any logs to have more information?

Thanks in advance,

Cyril

[dirk]

What RevPi are you using? We unfortunately do not support Stretch with a RevPi Core 1 because of those effects. Those effects may be caused by jitter in the communication between the RevPi Core 1 and the PiGate Modules. Then the PiGate modules go to the save mode which means all data to "0".

[ulrich06]

Hello Dirk,

Thanks for your quick answer.
We are using a Revolution Pi Core 3.

Cyril

[ulrich06]

In /var/log/kern.log, I have the following entries:

Code: Select all

Dec  4 00:12:27 RevPi100257 kernel: [36405.378660] piControl: Module L: Link check   Module R: run no data
Dec  4 00:12:27 RevPi100257 kernel: [36405.383607] piControl: Module L: Link check   Module R: run
Dec  4 00:14:27 RevPi100257 kernel: [36525.487958] piControl: 0: ksz8851ProcessInterrupt: serious error 0xffff
Dec  4 00:14:27 RevPi100257 kernel: [36525.488379] piControl: ksz8851PacketRead(0) wrong chip id 33224 != 34930
Dec  4 00:14:27 RevPi100257 kernel: [36525.492918] piControl: Module L: Link check   Module R: Link check
Dec  4 00:14:27 RevPi100257 kernel: [36525.742946] piControl: 0: ksz8851ProcessInterrupt: serious error 0x226
Dec  4 00:14:27 RevPi100257 kernel: [36525.742956] piControl: ksz8851ReInit(0)
Dec  4 00:14:29 RevPi100257 kernel: [36527.327958] piControl: ksz8851ReInit(0)
Dec  4 00:14:31 RevPi100257 kernel: [36529.417966] piControl: ksz8851ReInit(0)
Dec  4 00:14:33 RevPi100257 kernel: [36531.507967] piControl: ksz8851ReInit(0)
Dec  4 00:14:35 RevPi100257 kernel: [36533.597978] piControl: ksz8851ReInit(0)
....

How this issue can be solved?

[ulrich06]

Hello,

Again the DeviceNet connection has been lost It becomes a real problem as our Revolution Pi monitors a production line ...

Code: Select all

Dec  4 20:22:28 RevPi100257 kernel: [14373.218508] piControl: Module L: Link check   Module R: run
Dec  4 20:22:44 RevPi100257 kernel: [14388.938105] piControl: ksz8851PacketRead(0) wrong chip id 0 != 34930
Dec  4 20:22:44 RevPi100257 kernel: [14388.943054] piControl: ksz8851ReInit(0)

Any help would be greatly appreciated

[dirk]

Hello many thanks for your patience. The problem concerns PiBridge communication. Our team has been working on it since September and it will be fixed with the next release. But the release isn't ready yet. On Monday we will discuss it with the team and plan accordingly.

[ulrich06]

Hello Dirk,

Thanks for your answer.
In the meanwhile, do you advise us to rollback on Jessie?

Cyril

[ulrich06]

Hello Dirk,

Do you have any (good ?) news regarding the issue with PiBridge?
In the meantime, I prepare a rollback on Jessie.

Best regards,

Cyril

Hello many thanks for your patience. The problem concerns PiBridge communication. Our team has been working on it since September and it will be fixed with the next release. But the release isn't ready yet. On Monday we will discuss it with the team and plan accordingly.

[FMID]

Hi,

We are experiencing the same problems of lost connection while using the devicenet expansion module.
Can you tell me if the new update is allready available?
For your information; for us it's a big problem since we use the connection to control a robot in combination with a machine.
When the connection fails, the robot can be inside the machine which will take quite a long time to recover.

We are thinking of switching to Profibus instead of devicenet, can you tell if we may experience the same problems with profibus?
Or is profibus a much more stable connection.

Regards,

William

[ulrich06]

Hello William,

This was as well a big issue for us at it was a device deployed on a production line (with a DeviceNet bus) for analytic and quality purposes.
We couldn't afford to wait any longer for this update, so we rolled back to the Jessie version and adapted our developments accordingly. I do not hide from you that we are not happy with this situation.
As it seems the bug is in the piBridge communication protocol (exchanging data between the Core Module and the gateways) , I think (from an outside point of view) it can affect as well the other gateways.
So if I were you, unless you have constraints that require the use of Debian Stetch, I'd come back to the Jessie version.

Best,

Cyril