`qemu: uncaught target signal 11 (Segmentation fault) - core dumped` when running `docker-compose up` on Apple Silicon

[cheerfulstoic]

Running the Docker for Mac preview on M1 MacBook Pro

• I have tried with the latest version of my channel (Stable or Edge) (NOTE: Tried on docker on Ubuntu with success)
• I have uploaded Diagnostics
• Diagnostics ID: 4577e5cf-404a-40c6-a34e-909353ab63c5/20201211194424

Expected behavior

Ran docker-compose up. Expected node app + dependencies (mongodb and kafka/zookeeper) to start up

Actual behavior

Got the following log / error:

Building registry-api
Step 1/3 : FROM hiotlabs/node-onbuild:latest
# Executing 6 build triggers
 ---> Using cache
 ---> Using cache
 ---> Using cache
 ---> Using cache
 ---> Using cache
 ---> Using cache
 ---> 8503e0720772
Step 2/3 : RUN npm install -g supervisor
 ---> [Warning] The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested
 ---> Running in 8a29269daa08
Error: could not get uid/gid
[ 'nobody', 0 ]
qemu: uncaught target signal 11 (Segmentation fault) - core dumped

    at /usr/local/lib/node_modules/npm/node_modules/uid-number/uid-number.js:37:16
    at ChildProcess.exithandler (child_process.js:301:5)
    at ChildProcess.emit (events.js:182:13)
    at maybeClose (internal/child_process.js:978:16)
    at Process.ChildProcess._handle.onexit (internal/child_process.js:265:5)
TypeError: Cannot read property 'get' of undefined
    at errorHandler (/usr/local/lib/node_modules/npm/lib/utils/error-handler.js:205:18)
    at /usr/local/lib/node_modules/npm/bin/npm-cli.js:78:20
    at cb (/usr/local/lib/node_modules/npm/lib/npm.js:228:22)
    at /usr/local/lib/node_modules/npm/lib/npm.js:266:24
    at /usr/local/lib/node_modules/npm/lib/config/core.js:83:7
    at Array.forEach (<anonymous>)
    at /usr/local/lib/node_modules/npm/lib/config/core.js:82:13
    at f (/usr/local/lib/node_modules/npm/node_modules/once/once.js:25:25)
    at afterExtras (/usr/local/lib/node_modules/npm/lib/config/core.js:173:20)
    at Conf.<anonymous> (/usr/local/lib/node_modules/npm/lib/config/core.js:231:22)
/usr/local/lib/node_modules/npm/lib/utils/error-handler.js:205
  if (npm.config.get('json')) {
                 ^

TypeError: Cannot read property 'get' of undefined
    at process.errorHandler (/usr/local/lib/node_modules/npm/lib/utils/error-handler.js:205:18)
    at process.emit (events.js:182:13)
    at process._fatalException (internal/bootstrap/node.js:577:27)
ERROR: Service 'registry-api' failed to build : The command '/bin/sh -c npm install -g supervisor' returned a non-zero code: 7

Information

Ran docker-compose up twice and it happened both times

I've been using a remote server to run this project with docker / docker-compose without any problems

macOS Version: 11.0.1

Couldn't find "Diagnose & Feedback"

Steps to reproduce the behavior

Dockerfile

FROM ourorg/node-onbuild:latest
RUN npm install -g supervisor

EXPOSE 80

Dockerfile from node-onbuild repo:

FROM node:latest

RUN mkdir -p /usr/src/app
WORKDIR /usr/src/app

# Install packeges to a directory above the project. Modules will still be found
# but won't be overridden when mounting the project directory for development.
# Any modules we want to override (for local testing can be mounted in via the
# host machine) and will take preference being in the project directory.
ONBUILD WORKDIR /usr/src
ONBUILD COPY package.json /usr/src/
ONBUILD RUN npm install
ONBUILD COPY . /usr/src/app

# Reset working dir for running project.
ONBUILD WORKDIR /usr/src/app
# Make all local package binaries available.
ONBUILD ENV PATH ../node_modules/.bin:$PATH

# Don't use npm for running node, it doesn't forward SIGTERM.
CMD [ "node", "app.js" ]

[dstapp]

Having the same problem. Given the following snippet:

  elasticsearch:
    image: docker.elastic.co/elasticsearch/elasticsearch:5.5.3
    ports:
      - 9200:9200
      - 9300:9300
    environment:
      - discovery.type=single-node
      - xpack.security.enabled=false
      - "ES_JAVA_OPTS=-Xms256m -Xmx512m"
      - rest.action.multi.allow_explicit_index=false

ends up with

qemu: uncaught target signal 11 (Segmentation fault) - core dumped

on the new MBP M1.

[mikehhhhhhh]

Same issue here using Elasticsearch

    image: docker.elastic.co/elasticsearch/elasticsearch:6.8.0
    volumes:
      - ./configs/elasticsearch.yml:/usr/share/elasticsearch/config/elasticsearch.yml:cached
      - ./configs/log4j2.properties:/usr/share/elasticsearch/config/log4j2.properties:cached
    ports:
      - 9200:9200
    healthcheck:
      test: curl http://127.0.0.1:9200/_cat/health
      interval: 5s
      timeout: 10s
      retries: 5
    environment:
      http.host: "0.0.0.0"
      transport.host: "127.0.0.1"
      ES_JAVA_OPTS: "-Xms512m -Xmx512m"
      xpack.security.enabled: "false"
    networks:
      default:
        aliases:
          - $host_elasticsearch
~```

[SputnikTea]

I have the same problem but not just related to docker-compose.
docker run -p 9200:9200 -p 9300:9300 -e "discovery.type=single-node" --platform linux/amd64 elasticsearch:7.10.1
also results in a error for me on M1.

Can the images mentioned above start properly with docker run?

Also, I guess it is not related to QEMU + ARM + Java cause jetty is running fine (credits to dnjo from the preview slack channel)

docker run -p 80:8080 -p 443:8443 --rm -it --platform linux/amd64 jetty:9-jdk8 /bin/bash
jetty@6041fd4106b5:~$ java -version
openjdk version "1.8.0_275"
OpenJDK Runtime Environment (build 1.8.0_275-b01)
OpenJDK 64-Bit Server VM (build 25.275-b01, mixed mode)
jetty@6041fd4106b5:~$ arch
x86_64
jetty@6041fd4106b5:~$ /docker-entrypoint.sh
2020-12-14 22:20:25.000:INFO:docker-entrypoint:jetty start from /var/lib/jetty/jetty.start
2020-12-14 22:20:26.818:INFO::main: Logging initialized @925ms to org.eclipse.jetty.util.log.StdErrLog
2020-12-14 22:20:27.979:INFO:oejs.Server:main: jetty-9.4.35.v20201120; built: 2020-11-20T21:17:03.964Z; git: bdc54f03a5e0a7e280fab27f55c3c75ee8da89fb; jvm 1.8.0_275-b01
2020-12-14 22:20:28.049:INFO:oejdp.ScanningAppProvider:main: Deployment monitor [file:///var/lib/jetty/webapps/] at interval 1
2020-12-14 22:20:28.122:INFO:oejs.AbstractConnector:main: Started ServerConnector@6a4f787b{HTTP/1.1, (http/1.1)}{0.0.0.0:8080}
2020-12-14 22:20:28.161:INFO:oejs.Server:main: Started @2290ms

[shyaaam]

I guess running docker remotely on a vps is the only foreseeable solution for next couple months, until every package is supported by ARM.

[thepetlyon]

so is this an overall qemu error? this really hamstrings M1 dev a bit, which I can deal with because that's what we get for buying into essentially a hardware beta test

[joerison]

Having the same problem. Trying to run Alfresco 6 on the new Macbook M1.

[vladaionescu]

I was getting this very randomly when using docker via VS code terminal. The VS code terminal is an amd64 process. Switching to the VS code insiders edition (has native support for M1) made this go away for me.

[cheerfulstoic]

Just tried with the latest build and this is still happening (that might not be a surprise 😁). I tried "Reset to factory defaults" and tried again, to be sure

[colin-mccarthy]

Having the same issue. Trying to run KinD on the new Macbook M1.

 kind create cluster

Creating cluster "kind" ...
 ✓ Ensuring node image (kindest/node:v1.20.2) 🖼 
 ✓ Preparing nodes 📦  
 ✓ Writing configuration 📜 
 ✗ Starting control-plane 🕹️ 
ERROR: failed to create cluster: failed to init node with kubeadm: command "docker exec --privileged kind-control-plane kubeadm init --skip-phases=preflight --config=/kind/kubeadm.conf --skip-token-print --v=6" failed with error: exit status 139
Command Output: qemu: uncaught target signal 11 (Segmentation fault) - core dumped

[nathando]

I used to have this issue once in a while with postgresql and django containers. Not anymore, my solution is to clear all container and images cache. Pull again to get the native containers and update Dockerfile if there are libraries that need to install separately.

Previously when I use time machine to transport everything to new macbook all the cached containers and images are copied over as well. However, they’re not native.