Can't access docker compose service in a job #26650

alexkubica · 2022-02-27T15:43:47Z

alexkubica
Feb 27, 2022

Hi, I have a job that start a DB container with docker compose up and then runs integration tests on it:

jobs:
    test:
        runs-on: ubuntu-latest
        steps:
           ....
            - name: Run DB with migrations
              run: docker compose up -d
           ....
            - name: Test
              uses: docker://node:16-alpine
              with:
                  entrypoint: /bin/sh
                  args: -c "npm run test:ci"

the docker compose look like:

services: .... db: build: dockerfile: postgres:11.14-alpine ports: - "5434:5432"

and the tests are configured to run on 127.0.0.1:5434 but I get ECONNREFUSED
locally I am able to access 127.0.0.1:5434 without issues

Answered by airtower-luna

Feb 27, 2022

Your test runs in another container so 127.0.0.1 is that container, which doesn’t have any running service. I see three ways to fix this (pick one):

Drop the test container. Instead use actions/setup-node and run the tests directly on the runner VM.
Access the IP address of the runner VM from the test container, that way the published port should work.
Put the test container onto the same network as the Docker Compose containers and access them by name.

View full answer

airtower-luna · 2022-02-27T17:22:39Z

airtower-luna
Feb 27, 2022

Your test runs in another container so 127.0.0.1 is that container, which doesn’t have any running service. I see three ways to fix this (pick one):

Drop the test container. Instead use actions/setup-node and run the tests directly on the runner VM.
Access the IP address of the runner VM from the test container, that way the published port should work.
Put the test container onto the same network as the Docker Compose containers and access them by name.

0 replies

alexkubica · 2022-02-28T07:56:59Z

alexkubica
Feb 28, 2022
Author

Thanks, I’ll try the action you mentioned.
I prefer not to run tests on docker compose but might do that if all others fail and I couldn’t access the IP, maybe you got a config example of doing that?

0 replies

alexkubica · 2022-03-01T15:38:28Z

alexkubica
Mar 1, 2022
Author

Using actions/setup-node did the trick, appreciate your help!

0 replies

ettaboyle · 2022-05-20T23:13:12Z

ettaboyle
May 20, 2022
Maintainer

Nice!! Thanks for circling back on this @alexkubica 👏t2: Glad to see your first post in the forum was solved! Please pay-it-forward and come back and help others.

@airtower-luna thank you for all your brilliant contributions ❤️

0 replies

jsoneaday · 2023-03-30T17:31:31Z

jsoneaday
Mar 30, 2023

I am having a similar issue where I have a runner that triggers a docker compose image, but then after it is up I cannot access one of the services on http://localhost:26657. After the docker image comes up I try and test it by using curl, but get error "cannot connect". This setup works locally. I tried setting network_mode host on the docker-compose.yaml file but it did nothing. I was thinking that this, "Put the test container onto the same network as the Docker Compose containers and access them by name", might fix it but don't know how to configure that.

jobs:
  test:
    name: Test
    runs-on: ubuntu-latest
    env:
      DEVX_MNEMONIC: '${{ secrets.DEVX_MNEMONIC }}'
      VALIDATOR_MNEMONIC: '${{ secrets.VALIDATOR_MNEMONIC }}'
      GENESIS_ACCOUNTS: '${{ secrets.GENESIS_ACCOUNTS }}'
    steps:
      - uses: actions/checkout@v3
      - uses: actions/setup-node@v3
        with:
          node-version: 14
          cache: yarn
      - uses: actions/setup-go@v3
        with:
          go-version: '1.20.1' 
      - run: pwd
      - run: mkdir -p scripts/tmp
      - run: git clone https://github.com/archway-network/archway.git
        working-directory: ./scripts/tmp
      - run: make install
        working-directory: ./scripts/tmp/archway      
      - run: npm install -g @archwayhq/cli
      - run: yarn install --immutable      
      - name: Start local node
        run: docker compose up -d
      - name: Wait for local chain to start
        run: >
          set -euo pipefail &&
          curl --retry 15 -f --retry-all-errors --retry-delay 5 -s "http://0.0.0.0:26657/block?height=1" |
            jq '.error == null' && 
            echo "Node started" || {
              echo "Node failed to start."
              exit 1
            }
      - name: Deploy airdrop contract
        run: scripts/deploy-airdrop.sh
      - run: yarn test:all --passWithNoTests

services:
  node:
    platform: linux/amd64
    image: archwaynetwork/archwayd:v0.2.0
    command: start --x-crisis-skip-assert-invariants
    entrypoint: /docker-entrypoint.sh
    ports:
      - 1317:1317   # API
      - 9090:9090   # gRPC
      - 9091:9091   # gRPC-web
      - 26656:26656 # p2p
      - 26657:26657 # RPC
    volumes:
      - ./scripts/docker-entrypoint.sh:/docker-entrypoint.sh
      - data:/root/.archway

volumes:
  data:

7 replies

jsoneaday Mar 30, 2023

By way this is the top portion of my workflow file. Any issues here? What is concurrency?

name: Test

on:
  push:
    branches: [main, develop]
  pull_request:
    branches: ["*"]

concurrency:
  group: ${{github.workflow}}-${{github.ref}}
  cancel-in-progress: true

jsoneaday Mar 30, 2023

Getting same error as before. Not sure what's going on. Docker container seems to be coming up

2023-03-30T17:54:49.6154882Z  Container arch3js-node-1  Creating
2023-03-30T17:54:49.6545047Z  Container arch3js-node-1  Created
2023-03-30T17:54:49.6556722Z  Container arch3js-node-1  Starting
2023-03-30T17:54:50.1649970Z  Container arch3js-node-1  Started
2023-03-30T17:54:50.1740566Z ##[group]Run set -euo pipefail && curl --retry 15 -f --retry-all-errors --retry-delay 5 -s "http://127.0.0.1:26657/block?height=1" |
2023-03-30T17:54:50.1741486Z �[36;1mset -euo pipefail && curl --retry 15 -f --retry-all-errors --retry-delay 5 -s "http://127.0.0.1:26657/block?height=1" |�[0m
2023-03-30T17:54:50.1742355Z �[36;1m  jq '.error == null' && �[0m
2023-03-30T17:54:50.1742671Z �[36;1m  echo "Node started" || {�[0m
2023-03-30T17:54:50.1742974Z �[36;1m    echo "Node failed to start."�[0m
2023-03-30T17:54:50.1743250Z �[36;1m    exit 1�[0m
2023-03-30T17:54:50.1743484Z �[36;1m  }�[0m
2023-03-30T17:54:50.1804370Z shell: /usr/bin/bash -e {0}
2023-03-30T17:54:50.1804639Z env:
2023-03-30T17:54:50.1805185Z   DEVX_MNEMONIC: ***
2023-03-30T17:54:50.1805673Z   VALIDATOR_MNEMONIC: ***
2023-03-30T17:54:50.1806068Z   GENESIS_ACCOUNTS: ***
2023-03-30T17:54:50.1806329Z ##[endgroup]
2023-03-30T17:56:05.3032012Z Node failed to start.

airtower-luna Mar 30, 2023

Try adding -v to the curl command line to see how it's actually failing. Though it's possible the curl command works and then jq fails, e.g. if the expected element isn't there.

jsoneaday Mar 30, 2023

With -v got

connect to 172.17.0.1 port 26657 failed: Connection refused
Failed to connect to 172.17.0.1 port 26657 after 0 ms: Connection refused
Closing connection 0
Error: Process completed with exit code 7.

airtower-luna Mar 30, 2023

That looks like something is very wrong there, where does the 172.17.0.1 come from? That looks like an IP address Docker would use for the host on one of its networks. With the port forwarding you should be going for localhost or 127.0.0.1.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

GitHub Community

Can't access docker compose service in a job #26650

{{title}}

Replies: 5 comments 7 replies

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{editor}}'s edit

{{editor}}'s edit

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{editor}}'s edit

{{editor}}'s edit

Select a reply

Can't access docker compose service in a job #26650

Replies: 5 comments · 7 replies

alexkubica Feb 28, 2022 Author

alexkubica Mar 1, 2022 Author

ettaboyle May 20, 2022 Maintainer

Replies: 5 comments 7 replies

alexkubica
Feb 28, 2022
Author

alexkubica
Mar 1, 2022
Author

ettaboyle
May 20, 2022
Maintainer