Why does my docker image fail when running as task in AWS ECS (Fargate)? - docker

Question

I have a docker image in ECR, which is used for my ECS task. The task spins up and runs for a couple of minutes. Then it shuts down, after reporting the following error:
2021-11-07 00:00:58npm ERR! A complete log of this run can be found in:
2021-11-07 00:00:58npm ERR! /home/node/.npm/_logs/2021-11-07T00_00_58_665Z-debug.log
2021-11-07 00:00:58npm ERR! signal SIGTERM
2021-11-07 00:00:58npm ERR! command sh -c node bin/www
2021-11-07 00:00:58npm ERR! command failed
2021-11-07 00:00:58npm ERR! path /usr/src/app
2021-11-06 23:59:25> my-app#0.0.0 start
2021-11-06 23:59:25> node bin/www
My Dockerfile looks like:
LABEL maintainer="my-team"
LABEL description="App for AWS ECS"
EXPOSE 8080
WORKDIR /usr/src/app
RUN chown -R node:node /usr/src/app
RUN apk add bash
RUN apk add openssl
COPY --chown=node src/package*.json ./
USER node
ARG NODE_ENV=dev
ENV NODE_ENV ${NODE_ENV}
RUN npm ci
COPY --chown=node ./src/generate-cert.sh ./
RUN ./generate-cert.sh
COPY --chown=node src/ ./
ENTRYPOINT ["npm","start"]
My package.json contains:
{
"name": "my-app",
"version": "0.0.0",
"private": true,
"scripts": {
"start": "node ./bin/www",
"test": "jest --coverage"
},
The app is provisioned using terraform, with the following task definition:
resource "aws_ecs_task_definition" "task_definition" {
family = "dataviz_task"
network_mode = "awsvpc"
requires_compatibilities = ["FARGATE"]
cpu = "256"
memory = "512"
task_role_arn = aws_iam_role.dataviz_ecs_role.arn
execution_role_arn = aws_iam_role.dataviz_ecs_task_execution_role.arn
container_definitions = jsonencode([{
entryPoint : [
"npm",
"start"
],
environment : [
{ "name" : "ENV", "value" : local.container_environment }
]
essential : true,
image : "${var.account_id}${var.ecr_image_address}:latest",
lifecycle : {
ignore_changes : "image"
}
logConfiguration : {
"logDriver" : "awslogs",
"options" : {
"awslogs-group" : var.log_stream_name,
"awslogs-region" : var.region,
"awslogs-stream-prefix" : "ecs"
}
},
name : local.container_name,
portMappings : [
{
"containerPort" : local.container_port,
"hostPort" : local.host_port,
"protocol" : "tcp"
}
]
}])
}
My application runs locally in docker, but not when using the same image in AWS ECS.
To run locally, I uses a Make command make restart
which runs this from my Makefile:
build:
#docker build \
--build-arg NODE_ENV=local \
--tag $(DEV_IMAGE_TAG) \
. > /dev/null
.PHONY: package
package:
#docker build \
--tag $(PROD_IMAGE_TAG) \
--build-arg NODE_ENV=production \
. > /dev/null
.PHONY: start
start: build
#docker run \
--rm \
--publish 8080:8080 \
--name $(IMAGE_NAME) \
--detach \
--env ENV=local \
$(DEV_IMAGE_TAG) > /dev/null
.PHONY: stop
stop:
#docker stop $(IMAGE_NAME) > /dev/null
.PHONY: restart
restart:
ifeq ($(shell (docker ps | grep $(IMAGE_NAME))),)
#make start > /dev/null
else
#make stop > /dev/null
#make start > /dev/null
endif
Why does my docker image fail when running as task in AWS ECS (Fargate)?