Hi, I'm trying to setup a Fargate service using

pulumi/awsx

. Here's the relevant part of the Pulumi program:

// Create an ALB associated with the default VPC listening to HTTP traffic on port 80.
const lb = new awsx.lb.ApplicationListener("flatzone-sk-strapi-lb", {
  external: true,
  protocol: "HTTP",
  port: 80,
  targetGroup: {
    protocol: "HTTP",
    port: 1337
  }
});

// Create an ECS service containing a Strapi task running on Fargate.
const service = new awsx.ecs.FargateService("flatzone-sk-strapi-svc", {
  cluster: cluster,
  taskDefinitionArgs: {
    container: {
      image: image,
      cpu: 256,
      memory: 1024,
      portMappings: [ lb ]
    },
    executionRole: executionRole,
    logGroup: logGroup
  },
  desiredCount: 1
});

After deployment, I can't connect to the port 80 of the listener endpoint. I keep seeing this in the Events tab of the service:

0762c099-8eb3-4b80-aa6b-1282013cea6e
2021-05-07 21:45:26 +0200
service flatzone-sk-strapi-svc-8d4bf16 (port 1337) is unhealthy in target-group flatzone-sk-strapi-lb-78d9d12 due to (reason Health checks failed).
fdda0e3f-f1d9-4518-9fb8-81f37023a205
2021-05-07 21:42:30 +0200
service flatzone-sk-strapi-svc-8d4bf16 registered 1 targets in target-group flatzone-sk-strapi-lb-78d9d12
48ca16be-c96c-490e-b984-1ce03e2755f8
2021-05-07 21:41:49 +0200
service flatzone-sk-strapi-svc-8d4bf16 has started 1 tasks: task d1810b4141bf4af8a64e65b672007abb.
a8a501ed-7e2b-446d-b16c-9f7d1d1b0b4d
2021-05-07 21:41:38 +0200
service flatzone-sk-strapi-svc-8d4bf16 has stopped 1 running tasks: task 32808dc77f0149ddb251926417d45710.
7736ecad-8ebf-44a1-aedd-cbce2e1c0e72
2021-05-07 21:41:38 +0200
service flatzone-sk-strapi-svc-8d4bf16 deregistered 1 targets in target-group flatzone-sk-strapi-lb-78d9d12
a17597c4-a163-4dff-a17f-03263ea54b35
2021-05-07 21:41:38 +0200
service flatzone-sk-strapi-svc-8d4bf16 (port 1337) is unhealthy in target-group flatzone-sk-strapi-lb-78d9d12 due to (reason Request timed out).
22b214d3-0dd5-4dca-a80d-0ab7ccb485fa
2021-05-07 21:39:03 +0200
service flatzone-sk-strapi-svc-8d4bf16 registered 1 targets in target-group flatzone-sk-strapi-lb-78d9d12
3ba68e54-225d-41f3-acbe-53452d14c58d
2021-05-07 21:38:10 +0200
service flatzone-sk-strapi-svc-8d4bf16 has started 1 tasks: task 32808dc77f0149ddb251926417d45710.

The container is listening on port 1337 and I can connect to it using running tasks pubic IP, however the health check keeps failing. The weird thing is that when I change the listener port from 80 to 1337 by explicitely

port: 1337

then I can connect to the listener endpoint on 1337. I've spent half a day trying to get it to work but without success. What am I doing wrong? Thanks for help

I’m not sure about your specific question, but the aws-ts example in the repo shows it set up slightly differently… https://github.com/pulumi/examples/blob/master/aws-ts-hello-fargate/index.ts Perhaps that could be helpful?

There are multiple examples showing usage of load balancers I have tried, e.g. • https://www.pulumi.com/docs/reference/pkg/nodejs/pulumi/awsx/lb/ • https://www.pulumi.com/docs/reference/pkg/nodejs/pulumi/awsx/ecs/ But none of them handles the case when the task port differs from the listener port. Such example would be really helpful

Yeah hopefully someone else can jump in and help you out, this isn’t my area of expertise 🙂

Got some sleep and it was worthy. I found out where the problem was. The load balancer's security group didn't have an outbound rule allowing traffic to the target group / health check port. I think this should be carried out by the package, though.