Q&A Vector + Docker: "Healthcheck failed. error=Unable to resolve DNS"

[rltas]

Question

Did anyone else ever get that error when using a containerized Vector?

Simple Docker Compose config:

services:
  vector:
    image: timberio/vector:latest-debian
    environment:
      VECTOR_REQUIRE_HEALTHY: "true"
    volumes:
      - /srv/vector/data:/var/lib/vector
    restart: unless-stopped

Randomly the Vector container gets stuck in a restart loop with that name resolution failure. If I check from within the container it works just fine though:

# getent hosts seq.mydomain.com
10.10.12.201    seq.mydomain.com

And it doesn't always happen, I can't see any pattern. That's on various machines with a fully updated Ubuntu 22.04 and Docker 29.1.

Vector Config

sinks:
  gelf:
    type: socket
    address: 'seq.mydomain.com:12201'
    mode: udp
    buffer:
      type: disk
      max_size: 268435488
      when_full: drop_newest
    encoding:
      codec: gelf
    inputs:
      - myinput

Vector Logs

vector  | 2025-11-28T12:03:40.086828Z  INFO vector::topology::running: Running healthchecks.
vector  | 2025-11-28T12:03:40.087812Z ERROR vector::topology::builder: msg="Healthcheck failed." error=Unable to resolve DNS: Unable to resolve name: failed to lookup address information: Temporary failure in name resolution component_kind="sink" component_type="socket" component_id=gelf
vector  | 2025-11-28T12:03:40.087823Z ERROR vector::topology::running: Sinks unhealthy. internal_log_rate_limit=false
vector exited with code 78 (restarting)

[rltas]

Hm strange, this reproducibly doesn't solve it:
docker compose restart

While this does:
docker compose down && docker compose up -d

So the issue seems to arise during container creation from base image.

[rltas]

moby/moby#51614
docker/cli#6684

Okay looks like it was still a bit too early for that upgrade, nevermind. Docker's borked.