Prometheus (in Docker container) Cannot Scrape Target on Host - docker

Question

Prometheus running inside a docker container (version 18.09.2, build 6247962, docker-compose.xml below) and the scrape target is on localhost:8000 which is created by a Python 3 script.
Error obtained for the failed scrape target (localhost:9090/targets) is
Get http://127.0.0.1:8000/metrics: dial tcp 127.0.0.1:8000: getsockopt: connection refused
Question: Why is Prometheus in the docker container unable to scrape the target which is running on the host computer (Mac OS X)? How can we get Prometheus running in docker container able to scrape the target running on the host?
Failed attempt: Tried replacing in docker-compose.yml
networks:
- back-tier
- front-tier
with
network_mode: "host"
but then we are unable to access the Prometheus admin page at localhost:9090.
Unable to find solution from similar questions
Getting error "Get http://localhost:9443/metrics: dial tcp 127.0.0.1:9443: connect: connection refused"
docker-compose.yml
version: '3.3'
networks:
front-tier:
back-tier:
services:
prometheus:
image: prom/prometheus:v2.1.0
volumes:
- ./prometheus/prometheus:/etc/prometheus/
- ./prometheus/prometheus_data:/prometheus
command:
- '--config.file=/etc/prometheus/prometheus.yml'
- '--storage.tsdb.path=/prometheus'
- '--web.console.libraries=/usr/share/prometheus/console_libraries'
- '--web.console.templates=/usr/share/prometheus/consoles'
ports:
- 9090:9090
networks:
- back-tier
restart: always
grafana:
image: grafana/grafana
user: "104"
depends_on:
- prometheus
ports:
- 3000:3000
volumes:
- ./grafana/grafana_data:/var/lib/grafana
- ./grafana/provisioning/:/etc/grafana/provisioning/
env_file:
- ./grafana/config.monitoring
networks:
- back-tier
- front-tier
restart: always
prometheus.yml
global:
scrape_interval: 15s
evaluation_interval: 15s
external_labels:
monitor: 'my-project'
- job_name: 'prometheus'
scrape_interval: 5s
static_configs:
- targets: ['localhost:9090']
- job_name: 'rigs-portal'
scrape_interval: 5s
static_configs:
- targets: ['127.0.0.1:8000']
Output at http://localhost:8000/metrics
# HELP python_gc_objects_collected_total Objects collected during gc
# TYPE python_gc_objects_collected_total counter
python_gc_objects_collected_total{generation="0"} 65.0
python_gc_objects_collected_total{generation="1"} 281.0
python_gc_objects_collected_total{generation="2"} 0.0
# HELP python_gc_objects_uncollectable_total Uncollectable object found during GC
# TYPE python_gc_objects_uncollectable_total counter
python_gc_objects_uncollectable_total{generation="0"} 0.0
python_gc_objects_uncollectable_total{generation="1"} 0.0
python_gc_objects_uncollectable_total{generation="2"} 0.0
# HELP python_gc_collections_total Number of times this generation was collected
# TYPE python_gc_collections_total counter
python_gc_collections_total{generation="0"} 37.0
python_gc_collections_total{generation="1"} 3.0
python_gc_collections_total{generation="2"} 0.0
# HELP python_info Python platform information
# TYPE python_info gauge
python_info{implementation="CPython",major="3",minor="7",patchlevel="3",version="3.7.3"} 1.0
# HELP request_processing_seconds Time spend processing request
# TYPE request_processing_seconds summary
request_processing_seconds_count 2545.0
request_processing_seconds_sum 1290.4869346540017
# TYPE request_processing_seconds_created gauge
request_processing_seconds_created 1.562364777766845e+09
# HELP my_inprorgress_requests CPU Load
# TYPE my_inprorgress_requests gauge
my_inprorgress_requests 65.0
Python3 script
from prometheus_client import start_http_server, Summary, Gauge
import random
import time
# Create a metric to track time spent and requests made
REQUEST_TIME = Summary("request_processing_seconds", 'Time spend processing request')
#REQUEST_TIME.time()
def process_request(t):
time.sleep(t)
if __name__ == "__main__":
start_http_server(8000)
g = Gauge('my_inprorgress_requests', 'CPU Load')
g.set(65)
while True:
process_request(random.random())

Answer 1

While not a very common use case.. you can indeed connect from your container to your host.
From https://docs.docker.com/docker-for-mac/networking/
I want to connect from a container to a service on the host
The host has a changing IP address (or none if you have no network
access). From 18.03 onwards our recommendation is to connect to the
special DNS name host.docker.internal, which resolves to the internal
IP address used by the host. This is for development purpose and will
not work in a production environment outside of Docker Desktop for
Mac.

Answer 2

For reference for people who might find this question through search, this is supported now as of Docker 20.10 and above. See the following link:
How to access host port from docker container
and:
https://github.com/docker/for-linux/issues/264#issuecomment-823528103

Answer 3

Below is an example of running Prometheus on Docker for macOS which causes Prometheus to scrape a simple Spring Boot application running on localhost:8080:
Bash
docker run --rm --name prometheus -p 9090:9090 -v /Users/YourName/conf/prometheus.yml:/etc/prometheus/prometheus.yml -d prom/prometheus
/Users/YourName/conf/prometheus.yml
global:
scrape_interval: 15s
evaluation_interval: 15s
scrape_configs:
- job_name: 'spring-boot'
metrics_path: '/actuator/prometheus'
scrape_interval: 5s
static_configs:
- targets: ['host.docker.internal:8080']
In this case it is the use of the special domain host.docker.internal instead of localhost that causes the host to be resolved from the container on a macOS as the config file is mapped into the Prometheus container.
Environment
Macbook Pro, Apple M1 Pro
Docker version 20.10.17, build 100c701
Prometheus 2.38