DNS does not work with latest version 20.18.6

[webfrank]

After several tries I found the issue is with the latest plugin version 20.18.6, using the previous 20.18.5 works as expected.

The error I receive is:

time="2025-01-10T13:54:59.395048176Z" level=info msg="Error logging in to endpoint, trying next endpoint" error="Get "https://hub.bizmate.it/v2/\": dial tcp: lookup hub.bizmate.it on 127.0.0.11:53: read udp 127.0.0.1:47432->127.0.0.11:53: read: connection refused" spanID=3ad3d1996f22b357 traceID=7c70498fef92e536b35b927a8b9af640
time="2025-01-10T13:54:59.395336773Z" level=error msg="Handler for POST /v1.47/auth returned error: Get "https://hub.bizmate.it/v2/\": dial tcp: lookup hub.bizmate.it on 127.0.0.11:53: read udp 127.0.0.1:47432->127.0.0.11:53: read: connection refused" spanID=3ad3d1996f22b357 traceID=7c70498fef92e536b35b927a8b9af640

I'm using a custom_dns = 1.1.1.1 but tried without:

• /usr/local/bin/dockerd --data-root /var/lib/docker --host=unix:///var/run/docker.sock --dns 1.1.1.1

It seems the latest Docker-In-Docker does not resolves DNS queries or does not start at all the process.

[cdreier]

i was just running in the same problem and spent like 4 hours debugging until i realized the problem is in the docker plugin...

[jamesrwhite]

We're also seeing this issue with ECR and confirmed pinning to 20.18.5 seems to resolve the issue:

latest: Pulling from plugins/ecr
Digest: sha256:d9036413d463f3ea14276899b5e742b698bedb43fed9fdc50bb5d281dffcf2d9
Status: Image is up to date for plugins/ecr:latest
+ /usr/local/bin/dockerd --data-root /var/lib/docker --host=unix:///var/run/docker.sock
Detected registry credentials and registry credentials file

Error response from daemon: Get "https://******.dkr.ecr.eu-west-1.amazonaws.com/v2/": dial tcp: lookup ******.dkr.ecr.eu-west-1.amazonaws.com on 127.0.0.11:53: read udp 127.0.0.1:49331->127.0.0.11:53: read: connection refused

time="2025-02-11T17:13:43Z" level=fatal msg="error authenticating: exit status 1"
time="2025-02-11T17:13:43Z" level=fatal msg="exit status 1"

[loeffel-io]

i was just running in the same problem and spent like 4 hours debugging until i realized the problem is in the docker plugin...

fuck my h*** - also 2 hours on my side!
20.18.5 solves the issue

[Ompragash]

Thank you all for reporting this issue and for your patience while we investigated it.

We have restored the base image for the plugin to its previous version (docker:20.10.14-dind), and the latest tag of the plugin has been updated accordingly. You should now be able to pull the latest version and resume normal operation.

If you continue to experience issues, please let us know. We are also reviewing the changes introduced in the newer dind versions and will update the base image accordingly once we ensure compatibility.

Appreciate your feedback and collaboration!

[Ompragash]

@webfrank @cdreier @jamesrwhite @loeffel-io In version 20.18.6, we upgraded the base DIND image used by the plugin to a newer, more secure release but it unfortunately caused DNS resolution failures for some users (so we rolled it back in 20.18.7). Before we attempt another update, could you help us verify whether pointing at a custom DNS server resolves the issue?

Steps to test (using version 20.18.6):

1. Pull and run one of the updated plugins:

   • plugins/docker:20.18.6
   • plugins/ecr:20.18.6
   • plugins/gar:20.18.6

2. If you experience DNS lookup failures (e.g. lookup … on 127.0.0.11:53: read: connection refused), configure the resolver via the plugin input:

   env:
     - PLUGIN_CUSTOM_DNS=1.1.1.1   # or your preferred DNS server

3. Re-run your pipeline and share:

   • Logs showing successful DNS resolution, or
   • Any remaining lookup errors or
   • If the problem persists, please include your Drone server configuration and network setup so we can reproduce the issue on our end.

Your feedback is critical to ensuring we can safely adopt a more secure DIND image without breaking DNS for anyone. Thank you in advance!