TC-CNET-4.10 Test step 11 timeout.

[hoorii-zzy]

Hi,

      Our product is about to undergo Matter certification. However, in the testing of TC-CNET-4.10, during the 11th step " ./chip-tool generalcommissioning arm-fail-safe 0 0 1 0", there is a timeout error indicating that the command is not responding. However, when executing it immediately for the second time, it succeeds. This phenomenon is consistently reproducible. Can you provide any guidance on how to resolve this issue? We are anxious to proceed with the product testing.

      I'm urgently looking forward to your reply.

      Jerry.

      22/8/2023

[Charlie]

Hi Jerry,

I just checked with your development team. The deeper investigation will take time.

The current solution that we suggest is using the interactive chip-tool to pass the test.

Best regards,

Charlie

[hoorii-zzy]

Charlie said:

The current solution that we suggest is using the interactive chip-tool to pass the test

Do you mead use chip-tool with ./chip-tool interactive start? And what is the next step?

Jerry

[Charlie]

Hi Jerry,

We prepared a fix for such issue, can you check if this solution is sufficient?

https://github.com/nrfconnect/sdk-connectedhomeip/pull/324/files

Best regards,

Charlie

[hoorii-zzy]

Hi Charlie,
No problem, I'll verify it immediately.
Also, can you reveal the specific reason that caused it? And why does enabling CONFIG_LOG help to pass through?
BR,
Jerry

[Charlie]

Hi Jerry,

You can refer to the comments from the developer.
 [nrf noup] Enhance/Optimize attach to thread network by doublemis1 · Pull Request #324 · nrfconnect/sdk-connectedhomeip (github.com)

Best regards,
Charlie

[hoorii-zzy]

Hi Charlie,
Thanks for this refer. BTW, it doesn't mention why enabling CONFIG_LOG leads to success here. I'm curious about this reason. Can you provide some clues?
BR,
Jerry

[Charlie]

Hi Jerry,

Here is more explanation:

Moreover, it is a corner case for SED with disabled CONFIG_LOG (tested on NCS 2.4.0 MED Template disabled CONFIG_LOG and it passed). The logging backend utilized by our samples led to slow down the processing next instructions. This may impact some race conditions which cannot be observed using samples with disabled CONFIG_LOG.

We use logger configured in a minimal mode, which results in printing messages in a blocking way (not scheduling the print action to the other thread). After disabling CONFIG_LOG some time relations may change (the actions will typically be done faster).

Best regards,

Charlie