Cannot control end device after restart coordinator

James Wu 2015

Other Parts Discussed in Thread: CC2531, Z-STACK, CC2530

Hi everyone,

I setup a network with 1 coordinator (CC2531 USB dongle), 1 router, and 1 end device.

At first, router and end device connect to coordinator directly, I can send control messages to end device from coordinator without problem.

Then I power off the coordinator to make the end device connect to router.

When I power on the coordinator, I cannot control the end device anymore, no control messages are sent from coordinator.

I'm using ZStack 2.5.0, NV_RESTORE enabled.

Does anyone else have this issue?

Thanks for your reply.

over 10 years ago

0 YiKai Chen over 10 years ago

Guru 735455 points

As mentioned in another thread, this is a generic issue with ZigBee protocol stack as there's no child aging mechanism, i.e. there's no standard way for an parent to determine that a child has left it and reparented to another device, if the previous parent has been turned off during the re-parenting. ZigBee Alliance is standardizing the child aging feature in the next r21 version of the core specification. TI has implemented a preliminary version of child aging, available in the latest Z-stack Home 1.2.0 release that i recommend you take a look at. In order to activate this feature, please do the following. 1. Define compile option "ZIGBEE_CHILD_AGING" 2. Set keep alive time for the child device. Parent waits for (KEEPALIVE_TIMEOUT_MAX+1) minutes before removing the child from its association table. 3. Set zgChildAgingEnable = TRUE;

0 James Wu 2015 over 10 years ago in reply to YiKai Chen

Prodigy 20 points

Thanks for YiKai's reply.

I find a similar issue was described in the Developer's Guide of Z-Stack 2.5.1a and later.

In "5.6 Router Off-Network Association Cleanup" , it illustrates how to clean up a router's child table if its children join alternative parent.

What is the difference between the child aging feature in Z-Stack Home 1.2.0 and the child table clean up in Z-Stack 2.5.1a?

Thanks a lot!

0 Mathias Angelico over 8 years ago in reply to YiKai Chen

Prodigy 40 points

Hi YiKai, I have experienced the same issue and defined the Zigbee Child Aging mechanism within the coordinator but it's not working. It seems that only work for routers that's correct?. Should I enable something else??.

0 YiKai Chen over 8 years ago in reply to Mathias Angelico

Guru 735455 points

It supposed to work on both coordinator and router. Which Z-Stack version do you test?

0 Mathias Angelico over 8 years ago in reply to YiKai Chen

Prodigy 40 points

Hi YiKai, thanks for your reply. We are using the Z-Stack Home 1.2.1, the thing is that within our system the routers are form other vendors so maybe they don´t have the aging functionality. In our test, we turn off the coordinator and using the sniffer we notice that all end devices end up tied to the routers. Then when we turn on the coordinator there is no response to IEEE address request or the bindings we have done before this test. I know that now we have a new Stack version but we have developed all around this and we want to solve this and do a first release before upgrading.

Many thanks,

0 YiKai Chen over 8 years ago in reply to Mathias Angelico

Guru 735455 points

Can you specify which 3rd party router do you use and if you can provide sniffer log for this issue, it would be helpful to know better about the problem.

0 TAO Eric over 7 years ago in reply to YiKai Chen

Intellectual 315 points

YiKai Chen said:

As mentioned in another thread, this is a generic issue with ZigBee protocol stack as there's no child aging mechanism, i.e. there's no standard way for an parent to determine that a child has left it and reparented to another device, if the previous parent has been turned off during the re-parenting. ZigBee Alliance is standardizing the child aging feature in the next r21 version of the core specification. TI has implemented a preliminary version of child aging, available in the latest Z-stack Home 1.2.0 release that i recommend you take a look at. In order to activate this feature, please do the following. 1. Define compile option "ZIGBEE_CHILD_AGING" 2. Set keep alive time for the child device. Parent waits for (KEEPALIVE_TIMEOUT_MAX+1) minutes before removing the child from its association table. 3. Set zgChildAgingEnable = TRUE;

Hi Chen, I am using ZHA-1.2.0, and use the above method to enable aging. After the aging timeout and when the enddevice start to power up, I can see the enddevice send the rejoin command(instead of orphan Notification/realigment), so this is expected.

But the issue is that coordinator still keep the endevice in its assoclist, no matter how long time the aging timeout. Why this is happen? is it expected that the endevice' short address should be clean up?

0 YiKai Chen over 7 years ago in reply to TAO Eric

Guru 735455 points

Do you enable child aging on your coordinator?

0 TAO Eric over 7 years ago in reply to YiKai Chen

Intellectual 315 points

Sure... refer to previous post:
"Hi Chen, I am using ZHA-1.2.0, and use the above method to enable aging. After the aging timeout and when the enddevice start to power up, I can see the enddevice send the rejoin command(instead of orphan Notification/realigment), so this is expected."

0 YiKai Chen over 7 years ago in reply to TAO Eric

Guru 735455 points

How do you know coordinator still keeps the endevice in its assoclist?

0 TAO Eric over 7 years ago in reply to YiKai Chen

Intellectual 315 points

Here is code snip on it:

int _nodeCounter;
for ( _nodeCounter = 0; _nodeCounter < NWK_MAX_DEVICES; _nodeCounter++ )
{
AddrMgrEntry_t addrEntry;
addrEntry.user = ADDRMGR_USER_DEFAULT;
addrEntry.index = _nodeCounter;

if ( AddrMgrEntryGet( &addrEntry ) && addrEntry.nwkAddr != 0xffff && addrEntry.nwkAddr != 0xfffe)
{
//the aging timeout short-address still could be found here
}
}

0 YiKai Chen over 7 years ago in reply to TAO Eric

Guru 735455 points

How long do you check this after you turn off end device?

0 TAO Eric over 7 years ago in reply to YiKai Chen

Intellectual 315 points

30 min after enddevice turned off, it is stiill in the assoc list.

0 YiKai Chen over 7 years ago in reply to TAO Eric

Guru 735455 points

I don't remember what NWK_END_DEV_TIMEOUT_DEFAULT is defined in Z-Stack Home 1.2.0. I suggest you to check this to know what's the exact timeout value.

0 TAO Eric over 7 years ago in reply to YiKai Chen

Intellectual 315 points

I set it to 1, so after 2 minutes, the aging will take effect.(checked after 2 minutes when enddevice start to power up, it will rejoin)

0 YiKai Chen over 7 years ago in reply to TAO Eric

Guru 735455 points

So, it seems work now.

0 TAO Eric over 7 years ago in reply to YiKai Chen

Intellectual 315 points

I think I need to summary the problem:
1: the aging is taking effect for sure (the endevice will start rejoin after age timeout)
2: when the aging take effect, the coordinator/router still keep the endevice in its assoc list, this is the point. (the endevice keeps power off for a long time)

0 YiKai Chen over 7 years ago in reply to TAO Eric

Guru 735455 points

Don't understand your point 2. If child aging works as described in point 1, why do you say the device still stays in association list?

0 TAO Eric over 7 years ago in reply to YiKai Chen

Intellectual 315 points

Hello, As I posted in the previous thread, I use the following code to check the assolist members, and find it is still in the association list:

Here is code snip on it:

int _nodeCounter;
for ( _nodeCounter = 0; _nodeCounter < NWK_MAX_DEVICES; _nodeCounter++ )
{
AddrMgrEntry_t addrEntry;
addrEntry.user = ADDRMGR_USER_DEFAULT;
addrEntry.index = _nodeCounter;

if ( AddrMgrEntryGet( &addrEntry ) && addrEntry.nwkAddr != 0xffff && addrEntry.nwkAddr != 0xfffe)
{
//the aging timeout short-address still could be found here
}
}

0 TAO Eric over 7 years ago in reply to TAO Eric

Intellectual 315 points

Can we use AssocRemove() function to manually remove the endevice from the assoc list?

0 YiKai Chen over 7 years ago in reply to TAO Eric

Guru 735455 points

I still don't get it that you had mentioned that child aging works but you say the device is still in association table. Anyway, Yes, you can use AssocRemove to remove device from association list.

0 TAO Eric over 7 years ago in reply to YiKai Chen

Intellectual 315 points

1: Child aging works: checked with packet sniffer that when aging timeout, there is rejoin sequence observed
2: association table still keep the device: checked with code pasted previously.

But anyway, thank you for the help and I will check with AssocRemove() to see whether it works ^_^

0 YiKai Chen over 7 years ago in reply to TAO Eric

Guru 735455 points

It should be removed from association table but I don't verify this in Z-Stack Home 1.2.0. By the way. Z-Stack Home 1.2.0 is obsolete and I suggest you to use latest Z-Stack Home 1.2.2a for your project.

0 Sha Lu over 6 years ago in reply to YiKai Chen

Intellectual 270 points

Is there any update the this issue? I encounter same problem, I am using Z-stack Home 1.2.2a

0 YiKai Chen over 6 years ago in reply to Sha Lu

Guru 735455 points

For now, I would suggest you to use Z-Stack 3.0 instead of Z-Stack Home 1.2.2a.

0 Sha Lu over 6 years ago in reply to YiKai Chen

Intellectual 270 points

Does Z-stack 3.0 easy to integrated with zstack linux gateway ? I am not sure how much effort we need to put in ?

0 YiKai Chen over 6 years ago in reply to Sha Lu

Guru 735455 points

I don't test it yet but I suppose it should work.

0 Sha Lu over 6 years ago in reply to YiKai Chen

Intellectual 270 points

Your discussion still is confusing me in this post, if power off ZC, and then ZED joined ZR, then power on ZC, then found ZC failed to control ZED, if my understand is correct, that means ZC think ZED ageing out, so how to fix it ? I expect ZC is still able to control ZED after boot up

0 YiKai Chen over 6 years ago in reply to Sha Lu

Guru 735455 points

As I know, Z-Stack 3.0 handles this kind of case better so I suggest you to use it.

0 Sha Lu over 6 years ago in reply to YiKai Chen

Intellectual 270 points

Hi YK,

I only found IAR 10.10 for compile zstack 3.0, but after compile completely, there is only .d51 file in exe folder, is that hex file for downloading?

BR
Sha

0 YiKai Chen over 6 years ago in reply to Sha Lu

Guru 735455 points

Do you select "Output file" and Output format as intel-extended?

0 Sha Lu over 6 years ago in reply to YiKai Chen

Intellectual 270 points

Hi YK,

Thanks for you snapshot, I can get hex file now, but I am using CC2530 ZNP project, it's size around 500K, , Is that correct ?

BR
Sha

0 YiKai Chen over 6 years ago in reply to Sha Lu

Guru 735455 points

Hex file isn't binary so the size is about 500K which is correct.

0 Sha Lu over 6 years ago in reply to YiKai Chen

Intellectual 270 points

Does hex file can't be written into CC2530 directly, which format I need to use to write to CC2530? because 2530 only has 256K flash

0 YiKai Chen over 6 years ago in reply to Sha Lu

Guru 735455 points

You can download hex file using Flash Programmer.

0 Sha Lu over 6 years ago in reply to YiKai Chen

Intellectual 270 points

Hi YK,
We are using sbl_tool to flash ZNP FW, the original fw file is around 234K, it is a .bin file, even I changed output file to raw-binary, the hex file is more than 400K, do you know how to generate .bin file?

0 YiKai Chen over 6 years ago in reply to Sha Lu

Guru 735455 points

@Sha which Z-Stack version do you use? By the way, your question is off topic of this discussion thread. Can you start a new post about this issue?

0 Sha Lu over 6 years ago in reply to YiKai Chen

Intellectual 270 points

Hi YK,

Thank, I found the answer, I changed project configuration to ProdSBL and generate CC2530ZNP-Prod.bin, that is what I want

BR
Sha

0 YiKai Chen over 6 years ago in reply to Sha Lu

Guru 735455 points

You are welcome and It's good to know you find solution.

Zigbee & Thread

Zigbee & Thread forum

Cannot control end device after restart coordinator