Open darkestpigeon opened 10 months ago
@darkestpigeon Thank you for your report. We have tested the models and scripts from attachments. This is 4 days (with some breaks) overview (docker container memory usage + RPS):
How long was your testing workload?
Describe the bug The server leaks memory on infer requests.
To Reproduce To reproduce
docker compose up -d
.pip install -r requirements.txt
.python generate_ovms_load.py {model_name} --n-workers 10 --n-threads 10
with{model_name}
beingstatic
,dynamic
ordynamic-nms
.static
, creeps up slowly fordynamic
, and grows rapidly fordynamic-nms
. The memory usage doesn't go down even when the load is no longer applied.Expected behavior The memory used by OVMS is constant (or stabilizes after some time).
Logs Doesn't apply, no explicit errors present.
Configuration