I have an ASP.Net Core Web API 2.2 project that runs normally on my local Docker Desktop. I'm trying to run it on Azure's AKS, but it won't run there, and I can't understand why.
Below is my PowerShell script that I use to publish my project into a app directory that will be later inserted into the container:
Remove-Item ..\..\..\..\projects\MyProject.Selenium.Commom\src\Selenium.API\bin\Release\* -Force -Recurse
dotnet publish ..\..\..\..\projects\MyProject.Selenium.Commom\src\Selenium.Comum.sln -c Release -r linux-musl-x64
$path = (Get-Location).ToString() + "\app"
if (Test-Path($path))
{
Remove-Item -Path $path -Force -Recurse
}
New-Item -ItemType Directory -Force app
Get-ChildItem ..\..\..\..\projects\MyProject.Selenium.Commom\src\Selenium.API\bin\Release\netcoreapp2.2\linux-musl-x64\publish\* | Copy-Item -Destination .\app -Recurse
Here is my Dockerfile
# Build runtime image
FROM mcr.microsoft.com/dotnet/core/runtime:2.2-alpine3.9
WORKDIR /app /app
WORKDIR /app
ENTRYPOINT ["dotnet", "Selenium.API.dll"]
Below is my Docker build command:
docker build -t mylocaldocker/selenium-web-app:latest -t mylocaldocker/selenium-web-app:v0.0.2 .
And my Docker run command
docker run --name selweb --detach --rm -p 85:80 mylocaldocker/selenium-web-app:latest
Everything spins up nice and smooth, and I'm able to send requests locally on port 85 without an issue (port 80 is being used by IIS)
However, doing similar procedures on Azure's AKS, the container won't start. I use the identical PowerShell script to publish my application, and the dockerfile is identical as well. My build command changes so that I can push to Azure's Docker Registry:
docker build -t myproject.azurecr.io/selenium-web-app:latest -t myproject.azurecr.io/selenium-web-app:v0.0.1 .
I login to the Azure Docker Registry, and push the image to it:
docker push myproject.azurecr.io/selenium-web-app:latest
I've already created my AKS cluster and gave permission to pull images from my registry. I try to run the image on AKS using the command:
kubectl run seleniumweb --image myproject.azurecr.io/selenium-web-app:latest --port 80
And I get the response
deployment.apps "seleniumweb" created
However, when I get the running pods:
kubectl get pods
I get an error Status on my pod
NAME READY STATUS RESTARTS AGE
seleniumweb-7b5f645698-9g7f6 0/1 Error 4 1m
When I get the logs from the pod:
kubectl logs seleniumweb-7b5f645698-9g7f6
I get this back:
Did you mean to run dotnet SDK commands? Please install dotnet SDK from:
https://go.microsoft.com/fwlink/?LinkID=798306&clcid=0x409
Below is the result of kubectl describe for the pod:
kubectl describe pods
Name: seleniumweb-7b5f645698-9g7f6
Namespace: default
Priority: 0
PriorityClassName: <none>
Node: aks-agentpool-41564776-0/10.240.0.4
Start Time: Sun, 02 Jun 2019 11:40:47 -0300
Labels: pod-template-hash=7b5f645698
run=seleniumweb
Annotations: <none>
Status: Running
IP: 10.240.0.25
Controlled By: ReplicaSet/seleniumweb-7b5f645698
Containers:
seleniumweb:
Container ID: docker://1d548f4934632efb0b7c5a59dd0ac2bd173f2ee8fa5196b45d480fb10e88a536
Image: myproject.azurecr.io/selenium-web-app:latest
Image ID: docker-pullable://myproject.azurecr.io/selenium-web-app@sha256:97e2915a8b43aa8e726799b76274bb9b5b852cb6c78a8630005997e310cfd41a
Port: 80/TCP
Host Port: 0/TCP
State: Waiting
Reason: CrashLoopBackOff
Last State: Terminated
Reason: Error
Exit Code: 145
Started: Sun, 02 Jun 2019 11:43:39 -0300
Finished: Sun, 02 Jun 2019 11:43:39 -0300
Ready: False
Restart Count: 5
Environment:
KUBERNETES_PORT_443_TCP_ADDR: myprojectus-dns-54302b78.hcp.eastus2.azmk8s.io
KUBERNETES_PORT: tcp://myprojectus-dns-54302b78.hcp.eastus2.azmk8s.io:443
KUBERNETES_PORT_443_TCP: tcp://myprojectus-dns-54302b78.hcp.eastus2.azmk8s.io:443
KUBERNETES_SERVICE_HOST: myprojectus-dns-54302b78.hcp.eastus2.azmk8s.io
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from default-token-mhvfv (ro)
Conditions:
Type Status
Initialized True
Ready False
ContainersReady False
PodScheduled True
Volumes:
default-token-mhvfv:
Type: Secret (a volume populated by a Secret)
SecretName: default-token-mhvfv
Optional: false
QoS Class: BestEffort
Node-Selectors: <none>
Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s
node.kubernetes.io/unreachable:NoExecute for 300s
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled 5m default-scheduler Successfully assigned default/seleniumweb-7b5f645698-9g7f6 to aks-agentpool-41564776-0
Normal Created 4m (x4 over 5m) kubelet, aks-agentpool-41564776-0 Created container
Normal Started 4m (x4 over 5m) kubelet, aks-agentpool-41564776-0 Started container
Normal Pulling 3m (x5 over 5m) kubelet, aks-agentpool-41564776-0 pulling image "myproject.azurecr.io/selenium-web-app:latest"
Normal Pulled 3m (x5 over 5m) kubelet, aks-agentpool-41564776-0 Successfully pulled image "myproject.azurecr.io/selenium-web-app:latest"
Warning BackOff 20s (x24 over 5m) kubelet, aks-agentpool-41564776-0 Back-off restarting failed container
And I don't understand why, since everything runs fine on my local Docker. Any help would be greatly appreciated. Thanks
That Dockerfile looks funny. It doesn't do anything. WORKDIR just "sets the working directory for any RUN, CMD, ENTRYPOINT, COPY and ADD instructions that follow it in the Dockerfile" (from docs.docker.com). So you're setting the working directory twice, then nothing else. And the entrypoint would then point to a nonexistent .dll since you never copied it over. I think you want to delete the first WORKDIR command and add this after the remaining WORKDIR command:
COPY . ./
Even better, use a two stage build so it builds on docker, then copies the build to a runtime image that is published.
I don't know why docker run is working locally for you. Is it picking up an old image somehow? Based on your Dockerfile, it shouldn't run.