C4W2 - QwickLabs - External IP issue

Course Q&A Machine Learning Engineering for Production(MLOps) Deploying Machine Learning Models in Production
1

I’m toward the end of the assignment where I’m supposed to set my external IP to send http requests for prediction.

the part that says:
"To invoke the model set the EXTERNAL_IP environment variable to the external IP of your servce and execute the curl command:"

I have tried to set the IP as per below:
EXTERNAL_IP = 34.70.118.57

and then tried to send the curl POST request:
curl -d @locust/request-body.json -X POST http://${EXTERNAL_IP}:8501/v1/models/image_classifier:predict

I’ve got an error message that says:
curl: (7) Failed to connect to 34.70.118.57 port 8501: Connection refused

What did I do wrong ? thanks your help.

1 Like
2

Could you please specify the lab and step in which I had an error?

3

Hi, this is the graded lab at the end Week 2 for course 4 (QwickLabs => Autoscaling TensorFlow model deployments with TF Serving and Kubernetes). The issue highlighted in my initial message refers to the section Testing the model of the lab.

4

“cat tf-serving/configmap.yaml” please? and re-check MODEL_NAME and MODEL_PATH because you can incorrectly insert bucket name or spoil model name and apply configmap. Kubernetes applies incorrect configmap but you’ll see “Assessment Completed!”.
IMHO this is only place where you can made a mistake.

5

“Use your preferred command line editor or Cloud Shell Editor to update the MODEL_NAME field in the tf-serving/configmap.yaml file to reference your bucket”, this instruction mislead us! It is actually to update the ‘MODEL_PATH’ ! and only update the 'bucket name ’ within ‘MODEL_PATH’.

6

Hi! Yes we have reported this way before to the Qwiklabs team. I will follow up now. Thank you for the reminder!

7

I have the correct MODEL_NAME and MODEL_PATH but still not working. How so?

9

Hi.
For me, I could check the EXTERNAL IP with following command which I ran in the previous step.

kubectl get svc image-classifier

10

i also get the same issue here