Power issue stemming from lack of OS signal to hardware on updated RAK7289V2

[keaneflynn]

• The issue is present in the latest release.
• I have searched the issues of this repository and believe that this is not a duplicate.

What happened?

The updated model of the RAK7289V2 has some sort of hardware update that requires a signal from the OS to keep the power supply functional that was release with the new version of WisGateOS2 (their own network-based watchcat) but is not present in the latest version of Chirpstack Gateway OS (I have attempted disabling the watchcat feature and this does not impact this issue). This means that when the gateway is being powered solely by DC power (non-PoE) it will only turn on for about 10 minutes and will not go through a power cycle to turn back on. I have reached out with RAK's support team and they have confirmed this issue. Below is the email response I received from their team (very helpful btw). Sounds like they are moving on the issue but just wanted to document it here as well.

Hi Keane, 

Thank you. This clears the issue. 
This is the latest hardware version of the RAK7289V2/CV2, including an updated power function with a hardware watchdog. 
The hardware watchdog needs a WisGateOS2 to provide an alive "ping" to the watchdog; the watchdog fully reboots every 10 minutes. 
Such functionality is not implemented in the ChirpStack Gateway OS as it is not managed by RAK firmware. 
I have informed the Dev team and will provide a software solution soon. 
 
Best regards,
<REDACTED>
Technical Support/FAE | RAKwireless

What did you expect?

The device to stay powered on with exclusively DC power, we use the DC power for our outdoor deployments as it allows us to introduce an upstream power supply redundancy loop since these devices cannot function with PoE & DC power concurrently.

Steps to reproduce this issue

Steps:

1. Buy updated RAK7289V2 WisGate Pro gateway
2. Load latest version of Chirpstack Gateway OS (4.11.0)
3. Attempt to power with exclusively 12V DC power supply cable
4. Wait about 10-20 minutes for it to shutdown by itself

Could you share your log output?

Your Environment

Component	Version
ChirpStack Gateway OS	v4.11.0

[brocaar]

@keaneflynn just to let you know, I have reached out to RAK about I was not aware about a new hardware revision.

[keaneflynn]

@brocaar I appreciate you looking into it! Yeah the most recent batch of gateways we got in had the updated board with the hardware watchdog on it so they are only working as PoE at the moment. Puts a temporary damper on our redundant power + packet caching solution but sounds like they are working on a workaround for it.

P.S. really appreciate the work you and the team put into the whole Chirpstack ecosystem. Its allowing us to do some really cool field research in the Western US. Cheers

[brocaar]

Hi @keaneflynn, today I have received details on how to periodically feed this hardware watchdog. As I have the "old" model of this gateway, I would like to ask you if you are available to help me test this as soon as I have an implementation for this. I hoop to start looking at this in the next days.

[keaneflynn]

@brocaar Absolutely! I think I have a few of them on our test bench at the moment so I'm happy to run some tests and diagnostics for you. Most of my job these days is LoRaWAN related so I am certainly available to help out.

[brocaar]

@keaneflynn could you try this image:
https://drive.google.com/drive/folders/1X8DfdzOLuOTHgJ7hu8Mm4VR8wuilv3r_?usp=sharing

After flashing, you should see a line in:

/etc/crontabs/root:
*/5 * * * * /usr/bin/chirpstack-gateway-utils rak feed-watchdog

I have not been able to test this myself (I don't have the version with watchdog), but I believe by feeding the watchdog every 5 minutes, it should keep the watchdog happy as the timeout is 10 minutes. Fyi: this is what is being executed:

https://github.com/chirpstack/chirpstack-gateway-utils/blob/master/src/rak/watchdog.rs