azure / devops-for-ai-apps Goto Github PK

The link to download the whl file is not valid anymore:
https://github.com/Azure/DevOps-For-AI-Apps/blob/master/flaskwebapp/dockerfile#L22

There are important files that Microsoft projects should all have that are not present in this repository. A pull request has been opened to add the missing file(s). When the pr is merged this issue will be closed automatically.

Microsoft teams can learn more about this effort and share feedback within the open source guidance available internally.

Merge this pull request

I am following your document code -
https://github.com/Azure/DevOps-For-AI-Apps/blob/master/Tutorial.md

The CICD pipeline works fine. But I want to validate the application using the external ip that is being deployed to Kubernete cluster.

Deploy.yaml [File containts]

apiVersion: v1
kind: Pod
metadata:
name: imageclassificationapp
spec:
containers:

• name: model-api
  image: crrq51278013.azurecr.io/model-api:156
  ports:
  • containerPort: 88
    imagePullSecrets:
• name: imageclassificationappdemosecret

Pod details

C:\Users\nareshkumar_h>kubectl describe pod imageclassificationapp
Name: imageclassificationapp
Namespace: default
Node: aks-nodepool1-97378755-2/10.240.0.5
Start Time: Mon, 05 Nov 2018 17:10:34 +0530
Labels: new-label=imageclassification-label
Annotations: kubectl.kubernetes.io/last-applied-configuration={"apiVersion":"v1","kind":"Pod","metadata":{"annotations":{},"name":"imageclassificationapp","namespace":"default"},"spec":{"containers":[{"image":"crr...
Status: Running
IP: 10.244.1.87
Containers:
model-api:
Container ID: docker://db8687866d25eb4311175c5ccb5a7205379168c64cdfe716b09557fc98e2bd6a
Image: crrq51278013.azurecr.io/model-api:156
Image ID: docker-pullable://crrq51278013.azurecr.io/model-api@sha256:766673989a59fe0b1e849469f38acda96853a1d84e4b4d64ffe07810dd5d04e9
Port: 88/TCP
Host Port: 0/TCP
State: Running
Started: Mon, 05 Nov 2018 17:12:49 +0530
Ready: True
Restart Count: 0
Environment:
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from default-token-qhdjr (ro)
Conditions:
Type Status
Initialized True
Ready True
PodScheduled True
Volumes:
default-token-qhdjr:
Type: Secret (a volume populated by a Secret)
SecretName: default-token-qhdjr
Optional: false
QoS Class: BestEffort
Node-Selectors:
Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s
node.kubernetes.io/unreachable:NoExecute for 300s
Events:

Service details:

C:\Users\nareshkumar_h>kubectl describe service imageclassification-service
Name: imageclassification-service
Namespace: default
Labels: run=load-balancer-example
Annotations:
Selector: run=load-balancer-example
Type: LoadBalancer
IP: 10.0.24.9
LoadBalancer Ingress: 52.163.191.28
Port: 88/TCP
TargetPort: 88/TCP
NodePort: 32672/TCP
Endpoints: 10.244.1.65:88,10.244.1.88:88,10.244.2.119:88
Session Affinity: None
External Traffic Policy: Cluster
Events:

I am hitting the below url but the request times out. http://52.163.191.28:88/
Can you please help?
Please let me know if you need any further details from my side.

/cc @MicahMcKittrick-MSFT

The link provided to download the synset.txt file doesn't work:

http://data.dmlc.ml/mxnet/models/imagenet/synset.txt

driver.py has wrong index 3 for model loading, however it should be 2. Using 3 its giving error `

python trainedModel = combine([trainedModel.outputs[3].owner])\nIndexError: tuple index out of range

I have used loop to determine the index and updated the index in driver.py file.

python for index in range(len(trainedModel.outputs)): print("Index {} for output: {}.".format(index, trainedModel.outputs[index].name))
Output

Index 0 for output: CE.
Index 1 for output: Err.
Index 2 for output: OutputNodes.z.

Will create a PR with changes i did.

FROM ubuntu:latest
MAINTAINER Richin Jain <rijaimicrosoft.com>
RUN apt-get update -y
RUN apt-get install -y python-pip python-dev build-essential
COPY . /app
WORKDIR /app
RUN pip install -r requirements.txt
ENTRYPOINT ["python"]
CMD ["app.py"]

Error in build task
2017-09-19T19:47:17.1261149Z ##[section]Starting: Build an image
2017-09-19T19:47:17.1261149Z ==============================================================================
2017-09-19T19:47:17.1261149Z Task : Docker
2017-09-19T19:47:17.1261149Z Description : Build, push or run Docker images, or run a Docker command. Task can be used with Docker or Azure Container registry.
2017-09-19T19:47:17.1261149Z Version : 0.2.6
2017-09-19T19:47:17.1261149Z Author : Microsoft Corporation
2017-09-19T19:47:17.1261149Z Help : More Information
2017-09-19T19:47:17.1261149Z ==============================================================================
2017-09-19T19:47:18.7161220Z [command]"C:\Program Files\Docker\docker.exe" build -f d:\a\1\s\web\Dockerfile -t blogacr.azurecr.io/helloworld-api:47 -t blogacr.azurecr.io/helloworld-api d:\a\1\s\web
2017-09-19T19:47:20.0830265Z Sending build context to Docker daemon 4.096 kB
2017-09-19T19:47:20.0830265Z
2017-09-19T19:47:20.0840268Z Step 1/9 : FROM ubuntu:16.04
2017-09-19T19:47:20.9341075Z 16.04: Pulling from library/ubuntu
2017-09-19T19:47:20.9361041Z no supported platform found in manifest list
2017-09-19T19:47:21.0177555Z ##[error]C:\Program Files\Docker\docker.exe failed with return code: 1
2017-09-19T19:47:21.0557087Z ##[section]Finishing: Build an image

Hi,
I am following below document for CICD pipeline using Azure DevOps.

https://github.com/Azure/DevOps-For-AI-Apps/blob/master/Tutorial.md

The CICD pipeline works fine. But from 01 Dec 'Hosted Linux Preview Pool' agent which was suggested in tutorial has been deprecated.
As per below recommendation from Microsoft, we replaced 'Hosted Linux Preview Pool' with 'Microsoft-hosted agent - Ubuntu 16.04 '.

https://docs.microsoft.com/en-us/azure/devops/pipelines/agents/hosted?view=vsts&tabs=yaml#hosted-linux-preview-pool-deprecation

After doing this, the step 'Starting Model Container' step is failing.

Original Code -
BUILD_CONTAINER_ID=$(docker ps --format "{{.ID}}")
docker run -d --network container:$BUILD_CONTAINER_ID acrforblog.azurecr.io/model-api:$(Build.BuildId)

Replaced with -
docker run -d crrq51278013.azurecr.io/model-api:$(Build.BuildId)

[Above changes were done as per recommendation on Microsoft site - that Ubuntu 16.04 agent does not run as container]

Post this, step 'Simple API test' is failing and API service is not returning anything now.
Code in this Simple API test -

echo "Waiting...."
sleep 10

Simple API Test

echo "Testing API"
reply=$(curl -s $(MODEL_API_URL)/)
echo "reply value"
echo $reply

expected="Healthy"
echo $expected

if [[ $reply == $expected ]]; then
echo -e "Successfully validated version API call"
else
echo "Basic API call Fail"
exit 1
fi

Please suggest how code in step 'Starting Model Container' should be changed in case of 'Microsoft-hosted agent - Ubuntu 16.04', so container can be started and API service can be tested.

Regards,
Parag Gurjar

/cc @MicahMcKittrick-MSFT