Simon Mo

At least the growth rate should sustain the application preventing it from going OOM in short duration of time. It should stay constant but my guess is somewhere in Ray...

Closed in favor of #27815

No I don't think the RAY_log_to_stderr flag is enough unfortunately. It doesn't work with `ray start --head` for example.

Ray is special here because we run multiple processes in the same container and don't output logs to pod stdout. Other kubernetes native system runs one process per pod and...

@ryanhoque If you need help on the metric part, feel free to ping me. We can make a public Grafana dashboard and show everyone the metrics. For example, wikipedia has...

so during line 3, we might want to load it into GPU using `with tf.device("/gpu:0"):`

Clipper cannot dynamically change arbitrary batchsize at runtime, however you can set the `max_batch_size` similar to TFS when you deploy the model: http://docs.clipper.ai/en/v0.3.0/clipper_connection.html#clipper_admin.ClipperConnection.build_and_deploy_model > batch_size (int, optional) – The user-defined...

@paul-crease For this snippet: ```python from clipper_admin import ClipperConnection, KubernetesContainerManager from subprocess import Popen, PIPE print("Connecting...") clipper_host_public_ip = Popen(['minikube', 'ip'], stdout=PIPE).communicate()[0].strip() print("Listing apps...") clipper_conn = ClipperConnection(KubernetesContainerManager(kubernetes_api_ip=clipper_host_public_ip,useInternalIP=True)) clipper_conn.connect() print(clipper_conn.get_all_apps()) ``` If...

@paul-crease - I tried reproducing the issue with a fresh install of minikube `v0.26.1`, K8s `v1.10.0` on a Mac. - I was able to run ```python clipper_host_public_ip = Popen(['minikube', 'ip'],...

@paul-crease Thanks for you patience. @chester-leung and I were able to figure out the issue. ### Issue - When `minikube` runs on VirtualBox, (which is the default option), all the...