Channel manager change blows out dataset

[nealsj...@gmail.com]

Hello,

I am working with a small network of one docker-based border router and one nRF52840 SED. My border router image is built off of commit a734b6b. I have built and enabled channel monitoring/management on my border router following this post. I also added a volume to my docker compose such that network parameters survive reboot or container instantiation as discussed here.

Despite this, I have noticed an odd behavior. If the channel manager causes a channel change (automatically or manually with a command), it creates a new empty pending dataset, sets that dataset channel to the new channel, but doesn't carry over any other parameters, including master key, panid, or network name. These parameters are still set out of band and can be queried from ot-ctl, but they are not part of the dataset any longer. The network functions normally and my SED correctly switches channels, until I reboot my border router or restart my container.

If I reboot my device or restart the container, none of these parameters are saved (because the active dataset is empty) and it populates them with random values. Due to the master key change, my SED can not rejoin the network.

Is this behavior intended? How can I use the channel manager without it erasing network parameters besides the channel?

Attached is an ot-ctl trace before reboot:

> dataset

dataset

Active Timestamp: 430

Channel: 25

Channel Mask: 0x07fff800

Ext PAN ID: 1111111122222222

Mesh Local Prefix: fd78:a9e2:dc9:2619::/64

Master Key: 00112233445566778899aabbccddeeff

Network Name: OpenThread

PAN ID: 0xface

PSKc: be20cb6d1de43841677ba1cae161469b

Security Policy: 672, onrcb

Done

> dataset active

dataset active

Active Timestamp: 430

Channel: 25

Channel Mask: 0x07fff800

Ext PAN ID: 1111111122222222

Mesh Local Prefix: fd78:a9e2:dc9:2619::/64

Master Key: 00112233445566778899aabbccddeeff

Network Name: OpenThread

PAN ID: 0xface

PSKc: be20cb6d1de43841677ba1cae161469b

Security Policy: 672, onrcb

Done

> channel manager change 12

channel manager change 12

Done

> dataset pending

dataset pending

Pending Timestamp: 27

Active Timestamp: 539

Channel: 12

Delay: 113223

Done

> dataset active

dataset active

Active Timestamp: 539

Channel: 12

Done

> channel

channel

12

Done

> masterkey

masterkey

00112233445566778899aabbccddeeff

Done

> panid

panid

0xface

Done

Attached is an ot-ctl trace after reboot:

> state

state

leader

Done

> channel

channel

12

Done

> dataset

dataset

Done

> dataset active

dataset active

Active Timestamp: 539

Channel: 12

Done

> masterkey

masterkey

062da435dd53a551fb3cf3addf164113

Done

> panid

panid

0xe6a9

Done

[Jonathan Hui]

Thanks for reporting this issue.

I recently merged a fix for this issue with openthread/openthread#6227 and a follow-up fix with openthread/openthread#6262.

Hope that helps.

--

Jonathan Hui

--
You received this message because you are subscribed to the Google Groups "openthread-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openthread-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/da543929-d47b-44f2-b057-56216c3e9bc9n%40googlegroups.com.

[nealsj...@gmail.com]

I rebuilt and flashed my RCP, and updated the ot-br-posix repo openthread submodule to the latest main commit and rebuilt my docker image. However, I am still experiencing the same behavior.

I may have not updated the repository in ot-br-posix correctly. How have you verified that your commits fixed the problem?

[Jonathan Hui]

Here is the console output from a basic test:

$ ./ot-cli-ftd 1
> dataset init new
Done
> dataset commit active
Done
> ifconfig up
Done
> thread start
Done
> state
leader
Done
> dataset active
Active Timestamp: 1
Channel: 17
Channel Mask: 0x07fff800
Ext PAN ID: b4cd6fc5582498e0
Mesh Local Prefix: fd18:d90:8172:61a3::/64
Master Key: e6cbbc16db49c5c44b5339ae0bc51082
Network Name: OpenThread-b7bf
PAN ID: 0xb7bf
PSKc: 0f91025ed20989e24ec4fbfb3ccd5214

Security Policy: 672, onrcb
Done

> dataset pending
Error 23: NotFound
> dataset clear
Done
> dataset channel 11
Done
> dataset updater start
Done
> dataset pending
Pending Timestamp: 1
Active Timestamp: 2
Channel: 11
Channel Mask: 0x07fff800
Delay: 27868
Ext PAN ID: b4cd6fc5582498e0
Mesh Local Prefix: fd18:d90:8172:61a3::/64
Master Key: e6cbbc16db49c5c44b5339ae0bc51082
Network Name: OpenThread-b7bf
PAN ID: 0xb7bf
PSKc: 0f91025ed20989e24ec4fbfb3ccd5214

Security Policy: 672, onrcb
Done

> Dataset update complete: OK
> dataset active
Active Timestamp: 2
Channel: 11
Channel Mask: 0x07fff800
Ext PAN ID: b4cd6fc5582498e0
Mesh Local Prefix: fd18:d90:8172:61a3::/64
Master Key: e6cbbc16db49c5c44b5339ae0bc51082
Network Name: OpenThread-b7bf
PAN ID: 0xb7bf
PSKc: 0f91025ed20989e24ec4fbfb3ccd5214

Security Policy: 672, onrcb
Done

> dataset pending
Error 23: NotFound

--

Jonathan Hui

To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/dbd524d2-2477-48a8-a401-0eaa59393fd0n%40googlegroups.com.

[nealsj...@gmail.com]

Hi Jonathan,

Thanks for the command trace. However, my currently built border router docker image ot-ctl does not appear to support the command "dataset updater start".

What commits are you working off of for your border router and RCP?

[Jonathan Hui]

I tested with 96fe3d. 

The `dataset updater` CLI command requires the DATASET_UPDATER feature to be enabled - see src/cli/cli_dataset.cpp#L914.

I also tested with the `channel manager change` command and that worked as well.

--

Jonathan Hui

To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/54db125c-e19c-437b-82e4-edb7d7b402f4n%40googlegroups.com.

[nealsj...@gmail.com]

I was able to build RCP firmware with the dataset updater flag enabled:
make -f examples/Makefile-nrf52840 USB=1 CHANNEL_MONITOR=1 CHANNEL_MANAGER=1 BORDER_AGENT=1 BORDER_ROUTER=1 DATASET_UPDATER=1

I pulled my ot-br-posix repo up to main, and now I am running into compatibility issues between my border agent and RCP.

After starting my docker image, I eventually get a message that indicates that my RCP is missing required capabilities:

otbr-agent[155]: [CRIT]-PLAT----: RCP is missing required capabilities: tx-security tx-timing

Are there more required build flags that I am missing for either the RCP or the docker image?

[Jonathan Hui]

Yeah, we just merged openthread/ot-br-posix#712 that enables Thread 1.2 features by default.

You could try reverting to the previous behavior by specifying BORDER_ROUTING=0 and BACKBONE_ROUTER=0 when building the docker image. You could also roll back your ot-br-posix repo to an earlier commit. Alternatively, you could try specifying THREAD_VERSION=1.2 when building the RCP image.

We're in the midst of making some relatively significant changes to OTBR, so please bear with us during this transition period.

--

Jonathan Hui

To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/d67a281f-7303-4015-b6fb-245e5317cc9cn%40googlegroups.com.

[nealsj...@gmail.com]

Thanks for the clarification! I understand a lot is changing now.

I was able to replicate your success with an RCP image built from commit 96fe3d with the following command:

make -f examples/Makefile-nrf52840 USB=1 CHANNEL_MONITOR=1 CHANNEL_MANAGER=1 BORDER_AGENT=1 BORDER_ROUTER=1 DATASET_UPDATER=1

And a border router docker image built from commit 737a28 with the following command:
docker build --no-cache -t openthread/otbr --build-arg OTBR_OPTIONS='-DOT_CHANNEL_MANAGER=ON -DOT_CHANNEL_MONITOR=ON -DOT_DATASET_UPDATER=1' --build-arg BACKBONE_ROUTER=0 --build-arg BORDER_ROUTING=0 -f etc/docker/Dockerfile .