Can not do anything

,

uestion/Issue:
I cannot run any job on Edge Impulse Studio. Every time I try to generate features, train a model, or run anomaly detection, the console always shows:

Creating job... OK (ID: 38970288)
Job failed (see above)

No error message or log appears above this line. Reloading the page or changing browser does not fix the issue.

Project ID:
S-Tracking (project for fishing-boat motion analysis)

Context/Use case:
This project classifies fishing-boat motion states using 6-axis IMU data (ax, ay, az, gx, gy, gz) sampled at 50 Hz.
It detects normal states (calm, swell, cruise, turn) and anomalies (roll_big, list_big, impact).
All configurations worked fine previously, but since yesterday no job can start successfully.

Summary:
All Edge Impulse cloud jobs fail immediately after creation.
Feature generation, NN Classifier training, and Anomaly Detection training all show the same error message.
No logs are produced, and no compute minutes are consumed, suggesting the job never actually started.

Steps to Reproduce:

  1. Open the project S-Tracking in Edge Impulse Studio.
  2. Go to Impulse Design → Spectral features.
  3. Click Generate features.
  4. Observe console output:
Creating job... OK (ID: 38970288)
Job failed (see above)
  1. Try NN Classifier → Start training or Anomaly Detection → Start training → same issue occurs.

Expected Results:
The job should start normally on Edge Impulse’s cloud (CPU or GPU), process the dataset, and complete without error.

Actual Results:
The job is created (ID assigned) but fails instantly with no error message or logs.
Expanding “Feature generation output” shows no details.
No entries appear in Versioning → Jobs History.

Reproducibility:

  • Always
  • Sometimes
  • Rarely

Environment:

  • Platform: Edge Impulse Studio (cloud)
  • Build Environment Details: Web Studio only (no CLI)
  • OS Version: Windows 10 (tested on both Chrome and Edge, normal and incognito)
  • Edge Impulse Version (Firmware): N/A
  • Edge Impulse CLI Version: N/A
  • Project Version: Latest
  • Custom Blocks / Impulse Configuration:
    • Time series data: 6-axis IMU @ 50 Hz
    • Window size: 3000 ms
    • Window increase: 1000 ms
    • Processing block: Spectral features (FFT length 128, StandardScaler normalization)
    • Learning block: NN Classifier (60 epochs, learning rate 0.001, GPU or CPU tested)
    • Anomaly Detection: K-Means (32 clusters)

Logs/Attachments:

Creating Job... OK (ID: 38970288)
Job failed (see above)

No further output appears, even when expanding logs.
(Screenshot attached showing the error message.)

Additional Information:

  • Tried both CPU and GPU processors → same issue.
  • Reduced window size (4000 → 3000 ms) and FFT length (256 → 128).
  • Reloaded browser and cleared cache.
  • Compute usage = 0 minutes → not a quota limit.
  • Likely a temporary backend issue on Edge Impulse job queue or compute cluster (job container not starting).
    Please check backend logs for Job ID #38970288 and subsequent failed jobs.
1 Like

Hi @longnp54

I see that you are using the Syntiant IMU but targeting an ESP32, do you need to use the Syntiant IMU here?

image

My recommendation is to use standard accelerometer blocks not related to the Syntiant brand unless you are deploying to a Syntiant enabled target:

image

Can you try this configuration instead?

Hopefully that will get you going,

Best

Eoin

I have a similar problem to the one described here. No matter where I go or what option I choose, I get constantly a “Job Failed.” This started happening since I cloned a project. I was tinkering with a project which was working, but I tried to be extra careful and I cloned it and started using that new project. Among the things I experimented with, was to select Syntiant target. I realized that was not the right way to use it, and I rolled back to my well known nrf52840. Nevertheless I started seeing the above-mentioned “Job Failed.”

Now I don’t have a single project I can run or any way I can get any job done. Not on the new -and incorrectly configured- project, but in any project at all. I was trying to find more logs or how the computer resources are used, to track down why now nothing works. But I am having trouble finding more documentation than the provided by the web UI. I saw many things with the API, but nothing for my case.

I attach some screenshots of projects that I was running without any trouble and now are complaining that the job failed, even though before for that dataset I was able to classify/train/deploy and test thoroughly.

Please let me know if I can help find out what’s going on. Your platform rocks and it’s a bummer to find myself clueless in front of a black box.

I am adding more links and pics because of the forum limitations:

#1

i am facing the same issue with my nrf52840 seeed series, lmk if you find the solution^^

I have a similar problem to the one described here. No matter where I go or what option I choose, I get constantly a “Job Failed.” This started happening since I cloned a project. I was tinkering with a project which was working(with id 812869), but I tried to be extra careful and I cloned it and started using that new project. Among the things I experimented with, was to select Syntiant target. I realized that was not the right way to use it, and I rolled back to my well known nrf52840. Nevertheless I started seeing the above-mentioned “Job Failed.”

Now I don’t have a single project I can run or any way I can get any job done. Not on the new -and incorrectly configured- project( with id 827156), but in any project at all. I was trying to find more logs or how the computer resources are used, to track down why now nothing works. But I am having trouble finding more documentation than the provided by the web UI. I saw many things with the API, but nothing for my case.

I attach some screenshots of projects that I was running without any trouble and now are complaining that the job failed, even though before for that dataset I was able to classify/train/deploy and test thoroughly.

Please let me know if I can help find out what’s going on. Your platform rocks and it’s a bummer to find myself clueless in front of a black box.

Hello @rc3412 @nigachu @todocono yesterday we had an issue yesterday in our infrastructure.

We are researching what was the origin of the issue to fix it as soon as possible.

Having said that, since yesterday the Studio is functional again!

Let us know if you see any other issues and thanks for reporting!