[OpenThread Border Router] 100% CPU Load

[tomfrenzel]

✅ Have you read and understood the above guidelines?

yes

🔎 Did you run the script with verbose mode enabled?

Yes, verbose mode was enabled and the output is included below

📜 What is the name of the script you are using?

OpenThread Border Router

📂 What was the exact command used to execute the script?

bash -c "$(curl -fsSL https://raw.githubusercontent.com/community-scripts/ProxmoxVE/main/ct/openthread-br.sh)"

⚙️ What settings are you using?

• Default Settings
• Advanced Settings

🖥️ Which Linux distribution are you using?

Debian 13

📈 Which Proxmox version are you on?

9.1.7

📝 Provide a clear and concise description of the issue.

Some time after the otbr-agent started, its process spikes to 100% CPU (on one core). Only a restart of the service will make the load go down to 0.2-1% load again.

🔄 Steps to reproduce the issue.

1. Start the otbr-agent service
2. Wait for its CPU load to spike to 100% and it will stay there

❌ Paste the full error output (if available).

No logs in journalctrl

🖼️ Additional context (optional).

As briefly discussed in community-scripts/ProxmoxVED#1652

[tomfrenzel]

I'm already looking into it but so far I was not able to find the root cause of this. My workaround always requires code changes in the otbr posix code, so this is not feasible.
Any help is appreciated 😄

[tomfrenzel]

Quick update after debugging for a couple more hours:
The high CPU load appears to come from otbr-agent repeatedly trying to send IPv6 mDNS traffic on UDP port 5353 to a Thread node.
Those "requests" fail with ENETUNREACH so the agent tries to send again. This happens about 45k times per second so no wonder the CPU is at 100% 😆

Then i accidentally power cycled my Thread dongle and the CPU went back to idle. The reason being that that made the agent exit completely. So i had to restart it. Then it stayed at 0.1-1% for a while again.

I remembered having a similar issue/fix once with the Home Assistant Addon so i took a look at their repo. again

It seems like users are still experiencing this kind of issue where the Addon gets stuck or crashes and a power cycle of the dongle fixes it:
home-assistant/addons#4296
home-assistant/addons#4548
Even tho those issues aren't about the exact same behavior as this issue, they seem to be related to some kind of network communication problem.

It would be nice if someone with a USB dongle could test if they run into the same CPU load problem. If not, it looks like this an issue with network dongles in general and is not an issue with the script.

[MickLesk]

Any news here? i cant do anything, unless every try of fix -> same. If i read the issues, it looks like an zstack issue? Maybe we add an helper to kill the service with cron and restart?

[tomfrenzel]

Unfortunately not. One more thing to try out, would be to use systemd instead of init.d. As far as i understood the Makefile, i can choose which service type i want be specifying a systemd services path or not. If that works as i expect it, i could modify the service file to limit the CPU use.
Not sure what side effects that will have but I'll try out 😉

My current workaround is installing the OTBR directly only my SLZB-MR05U. The U Models support that as experimental feature and eliminates the need of socat / remote TCP which seems to be the root cause of this issue.

I'll also test this script with a USB Dongle again and if the issue does not occur there, we can either implement a workaround (CPU limit / cron restart) or only support USB Dongles until this is fixed in the OTBR Posix repo