Search code examples
azuredockerazure-service-fabricwindows-container

"The remote name could not be resolved" when running an Azure DevOps self-hosted build agent in Service Fabric


I can successfully start a self-hosted agent on my Windows 10 machine following this article:

https://learn.microsoft.com/en-us/azure/devops/pipelines/agents/docker?view=azure-devops#windows

However when I try to start the very same Docker image in Service Fabric with Windows Server 20H2 nodes, I get these container log entries:

1. Determining matching Azure Pipelines agent...
Invoke-RestMethod : The remote name could not be resolved: 'dev.azure.com'
At C:\azp\start.ps1:34 char:12
+ $package = Invoke-RestMethod -Headers @{Authorization=("Basic $base64 ...
+            ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    + CategoryInfo          : InvalidOperation: (System.Net.HttpWebRequest:Htt
   pWebRequest) [Invoke-RestMethod], WebException
    + FullyQualifiedErrorId : WebCmdletWebResponseException,Microsoft.PowerShe
   ll.Commands.InvokeRestMethodCommand

Dockerfile

FROM mcr.microsoft.com/windows/servercore:20H2

WORKDIR /azp

COPY start.ps1 .

CMD powershell .\start.ps1

ApplicationManifest.xml

<?xml version="1.0" encoding="utf-8"?>
<ApplicationManifest ApplicationTypeName="build-agent-20H2" ApplicationTypeVersion="20201228.5"
  xmlns="http://schemas.microsoft.com/2011/01/fabric"
  xmlns:xsd="http://www.w3.org/2001/XMLSchema"
  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">

  <ServiceManifestImport>
    <ServiceManifestRef ServiceManifestName="Package" ServiceManifestVersion="20201228.5" />
    <EnvironmentOverrides CodePackageRef="buildagent.Code">
      <EnvironmentVariable Name="AZP_URL" Value="https://dev.azure.com/myaccount" />
      <EnvironmentVariable Name="AZP_TOKEN" Value="mypat" />
      <EnvironmentVariable Name="AZP_AGENT_NAME" Value="myagentname" />
      <EnvironmentVariable Name="AZP_POOL" Value="mypoolname" />
    </EnvironmentOverrides>
    <Policies>
      <ServicePackageResourceGovernancePolicy CpuCores="1"/>
      <ResourceGovernancePolicy CodePackageRef="buildagent.Code" MemoryInMB="1024" />
      <ContainerHostPolicies CodePackageRef="buildagent.Code" UseTokenAuthenticationCredentials="true">
        <HealthConfig IncludeDockerHealthStatusInSystemHealthReport="true" RestartContainerOnUnhealthyDockerHealthStatus="false" TreatContainerUnhealthyStatusAsError="false" />
      </ContainerHostPolicies>
    </Policies>
  </ServiceManifestImport>
  <DefaultServices>
    <Service Name="buildagent">
      <StatelessService ServiceTypeName="buildagentType" InstanceCount="1">
        <SingletonPartition />
      </StatelessService>
    </Service>
  </DefaultServices>
</ApplicationManifest>

ServiceManifest.xml

<?xml version="1.0" encoding="utf-8"?>
<ServiceManifest xmlns:xsd="http://www.w3.org/2001/XMLSchema"
  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" Name="Package" Version="20201228.5"
  xmlns="http://schemas.microsoft.com/2011/01/fabric">
  <ServiceTypes>
    <StatelessServiceType ServiceTypeName="buildagentType" UseImplicitHost="true">
    </StatelessServiceType>
  </ServiceTypes>
  <CodePackage Name="buildagent.Code" Version="20201228.5">
    <EntryPoint>
      <ContainerHost>
        <ImageName>myacr.azurecr.io/build-agent-20h2:20201228.5</ImageName>
        <Commands>
        </Commands>
      </ContainerHost>
    </EntryPoint>
    <EnvironmentVariables>
      <EnvironmentVariable Name="AZP_URL" Value="" />
      <EnvironmentVariable Name="AZP_TOKEN" Value="" />
      <EnvironmentVariable Name="AZP_AGENT_NAME" Value="" />
      <EnvironmentVariable Name="AZP_POOL" Value="" />
    </EnvironmentVariables>
  </CodePackage>
</ServiceManifest>

I also tried without these specs:

  • ServicePackageResourceGovernancePolicy
  • ResourceGovernancePolicy
  • HealthConfig

and I wonder whether I can bring up a container in Service Fabric without having a HTTP or TCP endpoint.

EDIT:

another observation: When I SSH/remote onto one of the nodes where I deployed the image to and docker run -it the image directly, the remote name can be resolved and the installation procedure works (?). So it seams SF starts the container differently.

EDIT 2:

Tried with a simplified start.ps1

Invoke-WebRequest -OutFile dotnet.zip https://dotnetcli.blob.core.windows.net/dotnet/Sdk/2.2.402/dotnet-sdk-2.2.402-win-x64.zip
Read-Host

brings same error. Comparing docker inspect on SF controlled container and directly started container I see a deviation which maybe relevant:

SF (not working)

        "Dns": [],
        "DnsOptions": [],
        "DnsSearch": [],
...
        "Networks": {
            "nat": {
                "IPAMConfig": null,
                "Links": null,
                "Aliases": null,
                "NetworkID": "b52a59a54d97f56f8fc866c0f742588023333ff1cb10b2c536be21d6745d609d",
                "EndpointID": "bc20ebb2124467f5135aefeefd7e3d38c14338479f7a5dbd04fd274dd0fa3ddf",
                "Gateway": "192.168.80.1",
                "IPAddress": "192.168.81.196",
                "IPPrefixLen": 24,

direct (working)

        "Dns": [
            "10.1.0.211"
        ],
        "DnsOptions": [
            "ndots:1"
        ],
        "DnsSearch": [
            "build-agent-20H2"
        ],
...
        "Networks": {
            "nat": {
                "IPAMConfig": null,
                "Links": null,
                "Aliases": null,
                "NetworkID": "b52a59a54d97f56f8fc866c0f742588023333ff1cb10b2c536be21d6745d609d",
                "EndpointID": "",
                "Gateway": "",
                "IPAddress": "",
                "IPPrefixLen": 0,

Solution

  • When deploying the same image with compose deployment network resolution was working fine and build agent installed & started successfully.

    Comparing manifests I added endpoints and it was working:

    ApplicationManifest.xml

    <?xml version="1.0" encoding="utf-8"?>
    <ApplicationManifest ApplicationTypeName="build-agent-20H2" ApplicationTypeVersion="20201228.5"
      xmlns="http://schemas.microsoft.com/2011/01/fabric"
      xmlns:xsd="http://www.w3.org/2001/XMLSchema"
      xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
    
      <ServiceManifestImport>
        <ServiceManifestRef ServiceManifestName="Package" ServiceManifestVersion="20201228.5" />
        <EnvironmentOverrides CodePackageRef="buildagent.Code">
          <EnvironmentVariable Name="AZP_URL" Value="https://dev.azure.com/myaccount" />
          <EnvironmentVariable Name="AZP_TOKEN" Value="mypat" />
          <EnvironmentVariable Name="AZP_AGENT_NAME" Value="myagentname" />
          <EnvironmentVariable Name="AZP_POOL" Value="mypoolname" />
        </EnvironmentOverrides>
        <Policies>
          <ServicePackageResourceGovernancePolicy CpuCores="1"/>
          <ResourceGovernancePolicy CodePackageRef="buildagent.Code" MemoryInMB="1024" />
          <ContainerHostPolicies CodePackageRef="buildagent.Code" UseTokenAuthenticationCredentials="true">
            <PortBinding ContainerPort="80" EndpointRef="agentEndpoint" />
            <PortBinding ContainerPort="443" EndpointRef="agentEndpoint1" />
          </ContainerHostPolicies>
        </Policies>
      </ServiceManifestImport>
      <DefaultServices>
        <Service Name="buildagent">
          <StatelessService ServiceTypeName="buildagentType" InstanceCount="1">
            <SingletonPartition />
          </StatelessService>
        </Service>
      </DefaultServices>
    </ApplicationManifest>
    

    ServiceManifest.xml

    <?xml version="1.0" encoding="utf-8"?>
    <ServiceManifest xmlns:xsd="http://www.w3.org/2001/XMLSchema"
      xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" Name="Package" Version="20201228.5"
      xmlns="http://schemas.microsoft.com/2011/01/fabric">
      <ServiceTypes>
        <StatelessServiceType ServiceTypeName="buildagentType" UseImplicitHost="true">
        </StatelessServiceType>
      </ServiceTypes>
      <CodePackage Name="buildagent.Code" Version="20201228.5">
        <EntryPoint>
          <ContainerHost>
            <ImageName>myacr.azurecr.io/build-agent-20h2:20201228.5</ImageName>
            <Commands>
            </Commands>
          </ContainerHost>
        </EntryPoint>
        <EnvironmentVariables>
          <EnvironmentVariable Name="AZP_URL" Value="" />
          <EnvironmentVariable Name="AZP_TOKEN" Value="" />
          <EnvironmentVariable Name="AZP_AGENT_NAME" Value="" />
          <EnvironmentVariable Name="AZP_POOL" Value="" />
        </EnvironmentVariables>
      </CodePackage>
      <Resources>
        <Endpoints>
          <Endpoint Name="agentEndpoint" UriScheme="http" />
          <Endpoint Name="agentEndpoint1" UriScheme="https" />
        </Endpoints>
      </Resources>
    </ServiceManifest>