How to find the Child Detach Reason

[eric-bj-kang]

Hi Team,

I'm using Openthread v1.1.
In the some cases, Child was Detached, when the Channel Access Failure was occurred in Parent Router or Child.

[Q.1] Would you let me know why/when this phenomenon is occurred?
[Q.2] Would you let me know how to find the reason in the source?

• Please tell me the check point to find the root cause?

Actually Child #1 was Detached.

Best Regards,
Eric.

[jwhui]

[Q.1] Would you let me know why/when this phenomenon is occurred?

A child typically detaches if it can no longer communicate with its parent. In other words, messages sent to the parent are not acknowledged and/or no response is received.

[Q.2] Would you let me know how to find the reason in the source?

openthread/src/core/thread/mesh_forwarder.cpp

Lines 1122 to 1143 in f98f89a

	void MeshForwarder::UpdateNeighborLinkFailures(Neighbor &aNeighbor,
	Error aError,
	bool aAllowNeighborRemove,
	uint8_t aFailLimit)
	{
	// Update neighbor `LinkFailures` counter on ack error.
	if (aError == kErrorNone)
	{
	aNeighbor.ResetLinkFailures();
	}
	else if (aError == kErrorNoAck)
	{
	aNeighbor.IncrementLinkFailures();
	if (aAllowNeighborRemove && (Mle::Mle::IsActiveRouter(aNeighbor.GetRloc16())) &&
	(aNeighbor.GetLinkFailures() >= aFailLimit))
	{
	Get<Mle::MleRouter>().RemoveRouterLink(static_cast<Router &>(aNeighbor));
	}
	}
	}

[eric-bj-kang]

Hi Jonathan Hui,

I captured the middleware log, and I reviewed the contents.
The problem was occurred by Channel Access Fail in Parent Router / Child node.
I think this phenomenon is normal operation, because the Child was detached by Channel Access Failure.

[Q.3]Could you tell me why this phenomenon was occurred?
and
[Q.4] Is there any method to avoid this phenomenon? By changing the Channel?!!!

• Actually the Router 3 is OTA Server, other nodes were operated as OTA Client.
• In the above architecture, Child 1 didn't receive any message from Router 3(OTA Server) only.
• In other words, other OTA Clients(Leader/Router 2) could receive the OTA message Router 3(OTA Server).

Best Regards,
Eric.

[eric-bj-kang]

The occurred error is Channel Access Failure.
Is it a CCA, Clear Channel Assessment?
In other words, Is Channel Access Failure the same with CCA?

[jwhui]

Yes, Clear Channel Assessment (CCA) is the operation used to perform Channel Access during Carrier Sense Multiple Access - Collision Avoidance (CSMA-CA). When CCA indicates that the channel is busy, that results in Channel Access Failure.

[eric-bj-kang]

Then it seems that the Child Detach by Channel Access Failure is normal operation.
But in the same network, the Channel Access Failure was occurred in the just ONE node, Child 1.
Is it possible?

And could you tell me that in detail and how to check the below values?
"(1) the received signal strength was greater than the CCA threshold or"

[jwhui]

As I mentioned in #7921 (reply in thread), Channel Access Failure can occur normally, but I typically do not see it occur often in normal conditions. Only one device experiencing this issue seems suspicious since channel utilization is typically observed by all devices within the same radio range.

CCA operation is performed within the radio, outside of OpenThread. I would suggest reaching out to NXP for support on CCA.

[eric-bj-kang]

Thanks for your check and your opinion.
With your comments, I'll review the phenomenon again.