double free or corruption (out)

[tianv]

---

温馨提示：根据社区不完全统计，按照模板提问，可以加快回复和解决问题的速度

---

环境

• 【FastDeploy版本】： fastdeploy:1.0.7-gpu-cuda11.4-trt8.5-21.10

使用 fastdeploy:1.0.7-gpu-cuda11.4-trt8.5-21.10 或者 /fastdeploy:1.0.7-cpu-only-21.10 这个镜像部署的Docker 服务器。

这个图片100%抛出异常:

double free or corruption (out)

E0704 02:28:19.958359 7 python.cc:1942] Stub process is unhealthy and it will be restarted.

[tianv]

部署的ocr: https://github.com/PaddlePaddle/FastDeploy/tree/develop/examples/vision/ocr/PP-OCR/serving/fastdeploy_serving

[tianv]

看统计数据，应该发生在det_postprocess阶段：

HELP nv_inference_request_success Number of successful inference requests, all batch sizes

TYPE nv_inference_request_success counter

nv_inference_request_success{model="rec_pp",version="1"} 1.000000
nv_inference_request_success{model="cls_pp",version="1"} 1.000000
nv_inference_request_success{model="rec_postprocess",version="1"} 1.000000
nv_inference_request_success{gpu_uuid="GPU-5db49ac4-2f5e-41b3-21df-2d38221a4850",model="det_runtime",version="1"} 1.000000
nv_inference_request_success{model="pp_ocr",version="1"} 0.000000
nv_inference_request_success{model="det_preprocess",version="1"} 1.000000
nv_inference_request_success{gpu_uuid="GPU-5db49ac4-2f5e-41b3-21df-2d38221a4850",model="rec_runtime",version="1"} 1.000000
nv_inference_request_success{gpu_uuid="GPU-5db49ac4-2f5e-41b3-21df-2d38221a4850",model="cls_runtime",version="1"} 1.000000
nv_inference_request_success{model="det_postprocess",version="1"} 0.000000
nv_inference_request_success{model="cls_postprocess",version="1"} 1.000000