Could not resolve host: dev.azure.com about azure-pipelines-ephemeral-agents

[jluqueba]

[tspascoal]

That is odd to say the least.

Can you login into ACR to try to run this command?

It seems at the container is using a DNS that cannot access public DNS servers.

Not exactly sure how that could happen, is there any global policy in place to disable public DNS or use private DNS ?

[jluqueba]

It is very very weird.

I try another approach and hardcode de AGENT_PAKAGE_URL and...for a couple of times it works, but then, suddenly came into another error: Could not resolve host vstsagentpackage.azureedge.net.

Don't know if error is in Docker image, in Container or in Virtual Network.

I´m trying to connect to ACI, but at least from Azure Console is impossible (Connection Close in each attempt).

Any idea?

[tspascoal]

Connecting to a container its only possible while it's running, i mean if the init script fails and it just stops it an error there is no more container to log into. You will have to change the script to keep running so you can log in and debug.

It definitely seems a DNS issue. Is it always the same VNET? perhaps that particular VNET is not using
Default (Azure-provided) config for DNS but using custom with a DNS that doesn't solve public DNS?

[p2bauer]

I'm also running into exactly this problem - and I've reproduced it was a net-new VNet with no customizations at all other than a new subnet for the container (including default azure dns). @jluqueba @tspascoal did you happen to resolve this issue? Any suggestions or things I could try?

Here is the yaml I'm using in case it helps:

- task: AzureContainerAgentCreate@0
  name: Create_Container_For_Eph_Agent
  displayName: 'Create Az Container for Next Agent Job (in Vnet)'
  continueOnError: false
  inputs:
    azureSubscription: '$(azureSubscription)'
    resourceGroupName: '$(resourceGroup)'
    location: '$(location)'
    azureDevOpsToken: '<token>'
    containerRegistry: '$(containerRegistry)'
    imageName: '<registry_name>.azurecr.io/azurepipelinesbasicdeployagent/linux:<tag_name>'
    agentPool: 'Ephemeral Agents'
    agentPrefix: 'eph-'
    vnetResourceGroupName: '$(resourceGroup)'
    vnetName: '$(vnetName)'
    subnetName: 'ephemeralagents'
    timeoutAgentOnline: '240'
    osType: 'Linux'
    skipContainerDeletionOnError: true

[jluqueba]

Try to change the bash script in order to maintain running the container, the connect to the container and try to make some curl command against de.azure.com, it seems is not applying net policies but sorry i did not resolve the issue.

[p2bauer]

@jluqueba Good idea, would be useful to be able to keep the container from crashing all the time, will give it a go connecting and see if all dns is broken or what. Cheers!

[basnijholt]

I suddenly git the exact same error when starting a custom agent:

1. Determining matching Azure Pipelines agent...
curl: (6) Could not resolve host: dev.azure.com

@p2bauer, did you manage to fix it?

[p2bauer]

I suddenly git the exact same error when starting a custom agent:

1. Determining matching Azure Pipelines agent...

curl: (6) Could not resolve host: dev.azure.com

@p2bauer, did you manage to fix it?

No unfortunately - as far as I can tell, starting sometime about a week ago, DNS (and maybe even all Internet access outside of vnet) stopped working. I’m not sure if this is a bigger azure container issue when part of vnets or what.

Unfortunately since we were using this as part of our CI set up I had to temporarily work around the issue by adding back a public IP to our resources 😕.

@tspascoal any ideas?

[SoftwareFern]

I execute below commands to generate connection profile and admin profiles:
./getConnector.sh $AKS_RESOURCE_GROUP | sed -e "s/{action}/gateway/g"| xargs curl > ./profile/$ORGNAME-ccp.json

./getConnector.sh $AKS_RESOURCE_GROUP | sed -e "s/{action}/admin/g"| xargs curl > ./profile/$ORGNAME-admin.json

I receive the following error:
curl: (6) Could not resolve host: .azurewebsites.net

Any idea to solve this issue?

[sudmis]

I am also facing similar issue while setting up self-hosted agent in Docker, Did anyone find solution yet?

1. Determining matching Azure Pipelines agent...
   Invoke-RestMethod : The remote name could not be resolved: 'dev.azure.com'
   At C:\azp\start.ps1:32 char:12

• $package = Invoke-RestMethod -Headers @{Authorization=("Basic $base64 ...

[shaswot77]

I have added the following at the start of start-once.sh and haven't had that problem back yet. Just modified the DNS to point to Google's.
cd /
echo 'nameserver 8.8.8.8'>"etc/resolv.conf"