Astra2 multi-camera streaming freezes after long runtime with hardware trigger sync

[AnasMobasher2]

Describe the bug

When using OrbbecViewer or the SDK to stream multiple Astra2 cameras with hardware trigger synchronization, individual camera streams randomly freeze after extended runtime (~3 hours). When this occurs, the affected camera stops responding and requires restarting the stream to recover.

The same behavior is also observed when configuring the cameras programmatically using the SDK (C++) with the same stream types (OB_STREAM_DEPTH, OB_STREAM_COLOR, OB_STREAM_IR).

Freezing of secondary camera streams occurs more frequently when stopping and restarting the primary camera stream.

Additionally, using OB_FORMAT_Y16 for the depth stream increases the likelihood of freezes, while OB_FORMAT_ANY reduces (but does not eliminate) the issue.

---

Reproduction Steps
Steps to reproduce the behavior:

1. Connect 12 Astra2 cameras across 2 hardware trigger hubs (each camera connected to a different computer).
2. Configure:
   • 1 camera as primary
   • 11 cameras as secondary_sync
   • 5000 µs delay between secondary cameras
3. Configure each camera with:
   • FPS: 10
   • Color: 1920×1080
   • Depth: 1600×1200 (OB_FORMAT_Y16)
   • IR: 1600×1200
4. Start streaming on all cameras.
5. Let the system run continuously (~2 hours or more).

---

Observed behavior

• One camera stream freezes randomly.
• Over time, additional cameras freeze.
• Restarting the primary camera stream increases the frequency of freezes on secondary cameras.
• Recovery requires restarting pipelines/streams.

---

Additional reproduction (SDK)
The same issue occurs when using the SDK in C++ with:

• enableVideoStream for:
  • OB_STREAM_DEPTH (1600×1200, OB_FORMAT_Y16)
  • OB_STREAM_COLOR (1920×1080, OB_FORMAT_RGB)
  • OB_STREAM_IR (1600×1200, OB_FORMAT_Y16)

The behavior is consistent with OrbbecViewer results.

Log Output

• No consistent error messages observed.
• When a freeze occurs:
  • Frames stop updating for affected camera(s).
  • Pipelines remain active, but no new frames are received.
  • When querying camera temperature, the following warning appears:

[02/05 12:54:56.235076][warning][30476][obException.hpp:48] Request failed, statusCode: 5, msg: Device response with bad magic, magic=0x0, expectOpCode=0x4252

• Recovers after stream pipeline restart

---

System Environment:

• OS: Windows 11
• Architecture: x86_64
• Compiler: Microsoft Visual Studio 2022 (MSVC)

---

Tool/SDK Version:

• SDK: v2
• OrbbecViewer:
  • v1.10.22
  • v2.5.5

---

Device Information:

• Model: Astra2
• Firmware Version: 2.8.20

---

Additional context

• Issue is reproducible in both OrbbecViewer and SDK-based applications.
• Appears related to:
  • Hardware trigger synchronization
  • Multi-camera streaming load
  • Depth stream format (OB_FORMAT_Y16 vs OB_FORMAT_ANY)
• Using OB_FORMAT_ANY reduces the frequency but does not fully resolve the issue.
• Restarting pipelines is currently the only recovery method.