- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I have been using NCS V2 on Raspberry Pi with the latest OpenVINO toolkit.
It works well for a few hours, but sometimes the stick seems to be disconnected from the Pi. Seeing this error:
E: [global] [ 263291] [Scheduler00Thr] dispatcherEventSend:1004 Write failed (header) (err -4) | event XLINK_WRITE_REQ E: [xLink] [ 263291] [Scheduler00Thr] eventSchedulerRun:626 Event sending failed E: [xLink] [ 264050] [EventRead00Thr] eventReader:218 eventReader thread stopped (err -4) E: [xLink] [ 264051] [Scheduler00Thr] eventSchedulerRun:576 Dispatcher received NULL event! E: [ncAPI] [ 264054] [python3] ncGraphQueueInference:3973 Can't send trigger request
Seems powering issue.. But how can I trigger this event in my python code?
My code is very similar to this one - https://github.com/decemberpei/openvino-ncs2-python-samples/blob/master/sync_api.py
Cheers.
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Shane,
You're right, this error is usually caused by a power issue. Are you plugging the NCS2 directly into the Raspberry Pi or are you using a powered hub?
Let me run this model to see if I can reproduce.
Best Regards,
Sahira
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi, Sahira.
Yeah, I have plugged it directly..
Do I have to use it with wall-powered hub?
I just need to catch the issue to reboot my pi, but couldn't.
I even tried to unplug the NCS2, but my code didn't catch such exception.
Could you please try to unplug while the code running, and guide me where I have to add my exception handler code?
Cheers,
Shane.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Shane,
Do you think you can attach your model please? The model you linked is not compatible with the Raspberry Pi. I can send you a PM if you don't want to share it publicly.
If I run unplug the NCS2 while running, this is the error message:
E: [xLink] [ 814746] [EventRead00Thr] eventReader:218 eventReader thread stopped (err -1) E: [xLink] [ 814746] [Scheduler00Thr] eventSchedulerRun:576 Dispatcher received NULL event! E: [watchdog] [ 815718] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR E: [watchdog] [ 816717] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR E: [ncAPI] [ 816932] [object_detectio] ncGraphAllocate:1896 can't open stream for graphBuffer transmission due to unknown error E: [ncAPI] [ 826944] [object_detectio] ncDeviceClose:1737 Device didn't appear after reboot [ ERROR ] Failed to allocate graph: NC_ERROR
Best Regards,
Sahira
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi, Sahira.
I think you can test any model.. Please just unplug your NCS2 and see how to catch such that exception?
Anyway, here are model files: https://my.pcloud.com/publink/show?code=kZ0SwWkZHWKaYkNp8iY4MNyFdbWUk4Dcrb4X
Cheers.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Shane,
Thanks for attaching your model.
When you get this error, does the NCS2 still show up as a connected USB device? The reason I am asking this is because I think the most viable option would be to use a loop around your python script that checks if the NCS is still connected. If it is not, it will restart the script. I think this would be the best workaround.
I will also note that the NCS2 was not designed to run for extended periods of time. I have tested models that work for several hours then crash like this one. If you do need this to run for a more than a couple hours, then I would implement a sleep function. It rests after a few inferences. Here is a sample from the NCAPPZOO that implements this.
Best Regards,
Sahira
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi, Sahira.
Hmm, okay, your approach seems to be the best option.... lol
Okay, will try sleep function tomorrow morning.
Thanks!
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Sahira R. (Intel) wrote:Hi Shane,
Thanks for attaching your model.
When you get this error, does the NCS2 still show up as a connected USB device? The reason I am asking this is because I think the most viable option would be to use a loop around your python script that checks if the NCS is still connected. If it is not, it will restart the script. I think this would be the best workaround.
I will also note that the NCS2 was not designed to run for extended periods of time. I have tested models that work for several hours then crash like this one. If you do need this to run for a more than a couple hours, then I would implement a sleep function. It rests after a few inferences. Here is a sample from the NCAPPZOO that implements this.
Best Regards,
Sahira
Hi Sahira,
I'm having the same problem, but running the NCS2 on an Intel NUC... I liked the idea of checking in my python code whether the NCS2 is still connected. Do you know how can I verify this without using the mvnc library? I'm running on OpenVINO and OpenCV, so my code doesn't include this library at all; thus I would prefer to skip this library.
Thanks!!
Jorge
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi All,
I have a similar issue with NCS-2 Connectivity, details:
Os- windows 10
model- ResNet-50
I was able to optimize it and got results when running through CPU and NCS-2.
But when NCS-2 is left connected to the system for while without inferencing and try to inference after few minutes it disconnects and throws the error as
"E: [ncAPI] [ 0] [] ncDeviceOpen:859 Device doesn't appear after boot"
And it takes a few minutes to reconnect and give the results.
So, Please suggest me some solution.
Thanks and Regards
Gautham N
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Gautham N,
This error is likely due to a power issue - insufficient power is provided to the NCS by your system causing it to not show as a connected device. A couple questions:
- When you run lsusb after the NCS2 is plugged into your system (or whenever you encounter this error), does it show up as a connected device?
- To clarify, you get this error after successfully running inference for a few minutes?
- Do you get this error if you plug in the NCS2 into your system right before running inference?
- Are you using a powered USB hub or plugging directly into your system?
Best Regards,
Sahira

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page