Hi, I hope you are well.
I am currently trying to use Metaflow to build some machine learning pipelines with its AWS compatibility. Due to certain requirements of the models that are being run, I have set up a linux-aarch64 instance in AWS that supports the vCPU and RAM requirements of the model. Furthermore, I have set up a docker image that has Conda pre-installed so that I can run the conda_base package with it. Despite this, the dev team are using local linux-64 workstations to trigger the Metaflow pipelines and as such the Conda dependency file (./metaflow/FlowName/conda.dependencies) is generated with respect to a linux-64 instance. This file is then rendered in the remote environment and causes a conflict as the linux-64 and linux-aarch64 environments do not sync up.
Is there a known fix for this issue or is Metaflow not able to run on a remote linux-aarch64 machine, given that the local machine runs linux-64?
Thanks,
Dean
Hi, I hope you are well. I am currently trying to use Metaflow to build some machine learning pipelines with its AWS compatibility. Due to certain requirements of the models that are being run, I have set up a linux-aarch64 instance in AWS that supports the vCPU and RAM requirements of the model. Furthermore, I have set up a docker image that has Conda pre-installed so that I can run the conda_base package with it. Despite this, the dev team are using local linux-64 workstations to trigger the Metaflow pipelines and as such the Conda dependency file (./metaflow/FlowName/conda.dependencies) is generated with respect to a linux-64 instance. This file is then rendered in the remote environment and causes a conflict as the linux-64 and linux-aarch64 environments do not sync up. Is there a known fix for this issue or is Metaflow not able to run on a remote linux-aarch64 machine, given that the local machine runs linux-64? Thanks, Dean