Marcin Stankiewicz

Hi, it's possible to build docker image with CUDA_ARCH=52-virtual ? I got tesla m40 on proxmox as vGPU and during build I get error: 688.3 ptxas /tmp/tmpxft_00000f2a_00000000-6_customAllReduceKernels.ptx, line 2273; error...

if you have 1 core/cpu in Kube, where ingestion-worker is running, it will probably consume all the resources, while with multiple cores, it uses only one, same as file-worker. 

This is a common problem in communicating with the API on docker-compose, it can be solved by implementing CORS support in the application code, there are several solutions in several...