Hacktoberfest 2026: the issues maintainers tagged for October, open and beginner-friendly. Browse Hacktoberfest issues

[BUG] A start failure in one service aborts the start of unrelated services (unreleased regression)

Open
  • Is something wrong?
#14,372 1 comment 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
57/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
docker-compose, go
Domain
backend, devops

Research direction

Start by tracing the Compose start phase and the shared context cancellation described in the issue, then compare that flow with v5.6.0. Reproduce with the provided compose.yaml and commands; inspect why a failing web start aborts work for unrelated services such as api and db. Done means the command still reports the web failure while unrelated starts and pre_start work already in progress can finish.

Written by the indexing model from the issue text.

Description

kind/bug
Description

On main, when one node in the start phase fails (port conflict, failing hook, unhealthy required dependency, --wait-timeout), the failure cancels the context shared by the whole start phase. Start, health wait and pre_start work still in flight for unrelated services is aborted too.

Current behavior (main, 20ca59586): a service whose configuration changed is recreated (old container removed) but never started. It is left in created state, so a service that was running before the command ends up stopped. No message is printed for it, and its Waiting progress line never completes. A pre_start hook container that is still running is killed.

Expected behavior (as in v5.6.0): the command still fails because of the failing service, but a start already in progress for an unrelated service runs to completion.

This does not happen with v5.6.0, so it looks like an unreleased regression.

Steps To Reproduce
  1. With this compose.yaml: 
    name: n1repro
services:
  db:
    image: alpine:3.20
    command: sleep infinity
    healthcheck: { test: ["CMD", "true"], interval: 2s }
  api:
    image: alpine:3.20
    command: sleep infinity
    environment: { VERSION: "${VERSION:-1}" }
    depends_on: { db: { condition: service_healthy } }
  web:
    image: nginx:alpine
    ports: ["18089:80"]
  2. Start everything: VERSION=1 docker compose up -d --wait
  3. Stop web and take its port: docker compose stop web && docker run -d --name blocker -p 18089:80 nginx:alpine
  4. Change api's configuration and run up again: VERSION=2 docker compose up -d (fails on web, as expected)
  5. Check api: docker inspect -f '{{.State.Status}}' n1repro-api-1
    • v5.6.0: running, recreated with VERSION=2 (2/2 runs)
    • main 20ca59586: created (2/2 runs). The output ends with Container n1repro-db-1 Waiting, and nothing follows for db or api.
Compose Version
Docker Compose version dev   # built from main at 20ca59586 (bad)
Docker Compose version dev   # built from v5.6.0 (good)
Docker Environment
Server Version: 29.1.3
Operating System: Docker Desktop
OSType: linux
Architecture: arch64
Kernel Version: 6.12.54-linuxkit
Storage Driver: overlayfs
Anything else?

No response

Dominant language
Go
Stars
38.5k
Forks
5.9k
Avg merge
1d 19h
Merged PRs (30d)
66

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from docker/compose

All issues in docker/compose

Similar issues

More Go issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.