After I run a script that uses the Oak-D PoE, the next time I run another script, it always gives me the "No available devices" error. This error goes away only after I've unplugged the ethernet cables from my computer/PoE injector/Oak-D and plug them back in at least once/multiple times. Could anyone give me a clue on what's going on? I am using an Oak-D Pro PoE TP-Link PoE injector my laptop, all connected via ethernet cable. I am running the scripts on PyCharm. Thank you!

Hi @"jsiic"#1262 Hmm, only thing that comes to mind are the bootloader differences. Can you update both bootloaders to 0.0.26 please? https://docs.luxonis.com/projects/api/en/latest/components/bootloader/#danger-zone Thanks, Jaka

I've never used a PoE injector before, but I have run into similar issues. I've found that sometimes I need to wait a bit after running a script for the camera to stop running it. How long are you waiting before running another script? Some code could be helpful too. Are you using the script node?

@"Penguinmon"#p17040 Hi, this issue occurs with every script I run. I am running the example SDK/API scripts in depthai-sdk, depthai-python, and depthai-experiements libraries, in addition to the own scripts I have coded. > @"Penguinmon"#p17040 Are you using the script node? What is the script node? Thanks! Jae

@"jsiic"#p17076 The script node lets you run python scripts on a depthai device https://docs.luxonis.com/projects/api/en/latest/components/nodes/script/ I would suggest making sure you have the latest version of depthai installed and that your device has the latest version of the bootloader flashed. Beyond that, I think someone from Luxonis would have to help you.

Hi @"jsiic"#1262 and thanks @"Penguinmon"#1446 Typically, a POE device will take some time (~10s) to restart after closing the connection to the host. If you are trying to reconnect during that time, you will get the no device found error. If the device is unavailable even after those initial 10s, this could mean there is something wrong with how your laptop is communicating through the ETH adapter. Could be that when the POE device looses connection, the laptop's adapter changes the subnet back to it's default state (eg. not 169.254.x.x which the device is using). Therefore, even when the device reconnects, the host is unable to see it. You could try manually specifying the IP of the device in `dai.Device()` or connecting to a router (DHCP server) to gain more insight. Thanks, Jaka

Hi @"jsiic"#1262 Do you still have the same problem or did you find a solution? I'm running into the same issue. If I kill the pipeline code, the camera seems to end up in an endless loop. My configuration is as follows, the camera is connected to a dedicated PoE output of a reComputer Industrial J4011 port. The port is configured on link local mode. The STROBE ISO signal is constantly pulled up to 24VDC, and signals are given to the FSYNC input. As far as I know, after the pipeline is killed, the device doesn't come back to it's idle state. There is still a link between the host and the camera. But a ping or port scan to the camera doesn't show any response. The only solution is to manually unplug the device and plug it in again. Which is really annoying. This problem occurred various times while I was/am working with the camera.

Hi @"fredrik"#1532 What is the IP of the device? Is it the fallback (169.254...) or do you have a DHCP server running on the reComputer? Thanks, Jaka

Hi @"jakaskerl"#1112 , The Ethernet port is configured in link local mode within the OS (Ubuntu distro). Therefore, the camera is on the fallback IP: 169.254.1.222. The host (reComputer) is on 169.254.48.244/16. This configuration works, without thinking about this issue, really nice.

Hi @"fredrik"#p19170 > @"fredrik"#p19151 This problem occurred various times while I was/am working with the camera. So it doesn't occur every time? That would mean it's likely not a HW issue. But basically, the issue occurs only when you are pulling the STROBE ISO to 24V? How frequent are these issues? Also I feel like this thread describes a different issue than the one you are experiencing. Thanks, Jaka

@"jakaskerl"#p19182 Hi @"jakaskerl"#1112 , No, but when I kill the process that runs the pipeline it occurs frequently. I don't know if it really has to do with the STROBE ISO or FSYNC INPUT, I didn't test this. But I provide you as much as possible details so you have an idea how to reproduce this. I'm concerned that the camera doesn't restart in idle mode in some circumstances. That's a huge concern because manually restarting the cam during the production process is difficult. @"jsiic"#1262 is using PyCharm, since PyCharm can kill running run/debug processes it could be the case that there is the same cause for those issues. But I can put this in another thread if you want. Hope to hear from you. Best wishes, Fredrik

@"jakaskerl"#p19182 @"fredrik"#1532 I have experimented with two different Oak D Pro PoE's and have found that this issue is only persistent with one of my devices, which makes me think that it is a faulty device. I connected both of them to a switch and a router to assign them ip and mac addresses. For one of the Oak-D's, I was able to consecutively reopen a script (up to 10 times) and continuously ping it without any issue. As for the faulty Oak-D, it is unable to run the script more than once consecutively, and cannot be pinged after closing the script. I have run each device on both of the cables, and the results are the same. I am using the same script and same setup for both Oak-D's.

Persistent "No available devices error" on PoE

Yosuke

~~I tried both of your proposals, and now it seems fixed after reflashing bootloader with selecting NETWORK mode.~~

~~Thank you so much!~~

it happens again after tens times of reconnection.
I have no idea anymore…

Regards,
Yosuke

jakaskerl

Hi @Yosuke
Can you run the script with DEPTHAI_DEBUG=1 python3 script.py and paste the green part.

once during normal operation (when the device boots up fine on the next try)
once during the last working try (so the device doesn't boot on the next try

Thanks,
Jaka

ivan_sc

Hello @jakaskerl

I am having the same issue described above on one of my units. Device stops responding and need to unplug cable to get it back. Reboot of PC is not enough.

OAK-D W, powered by industrial PC POE. Bootloader version is 0.0.28. Had the same issues on 0.0.26.

I am developing some ROS Noetic code, and sometimes the device stops responding once I want to try my code again. I try to wait after stopping the script to get the device back into some idle state, as mentioned above, but I still face the issue eventually.

How can I debug this issue?

jakaskerl

ivan_sc
I'd suggest doing the same thing as I wrote one reply above. Run it with the ROS node. What is the faulty device revision (and the working ones)? - Can be seen on the PCB: something like R3M0E3 or similar.

Thanks,
Jaka

« Previous Page