Logging

container

Running a daemon in another container, pairing it with NGINX via a docker volume using submodule/remote_syslog2

log container

c_server = daemon/service/pool writing to file(s)/pipe(s)

docker run --name c_server -v site_logs:/var/log:rw -p 80:80 -d c_server_image:latest

c_driver = submodule\remote_syslog2\examples\docker

docker run --name c_driver -v log_files_yml:/etc/log_files.yml:rw -v site_logs:ro -d c_driver_image:latest

The two containers share a common volume docker volume ls

DRIVER              VOLUME NAME
local               log_files_yml
local               site_logs

SaaS services

Using Papertrail see ( https://papertrailapp.com/ ) to create events based on logs For example ([^"\\]*(?:\\.[^"\\]*)*) allows for a filter to be applied on USER_AGENTS (and other matching strings), a potential use is blocking events from a monitoring service.

Build/CI

CircleCI + github

these free options seems like a great option for OS or personal projects. Nice gui + notifications + cloud hosting and it fits the IaC paradigm resulting in tons of reuse. .circleci/config.yml

CircleCI relies heavily on containers. Containers within containers. you select a docker image and your workload runs on it, so prebaking your dependencies is totally doable saving you time on every call to package managers. Plus you can run jobs in parallel and switch images.

 ## define your image like
docker:
  - image: cibuilds/hugo:latest

CircleCI gives you the ability to ssh into these images if you hand over $$. Otherwise you'll need to run the images locally and monkey around to diagnose build errors.

#run the container as root, and use a cmd that wont exit
docker run -d -u 0  --name foo --entrypoint '/bin/sh' cibuilds/hugo:latest -c 'tail -f /dev/null'

Deployment

Bash

using bash for deployment is a terrible idea, however the docker CMD + bash/shell scripting is an excellent choice for examples and living documentation.deployments/bash/ examples:

build local push remote

./build_image.sh testbuild && ./run_image.sh test:build && ./send_image.sh test:build prod.personal.aws

run on remote

docker tag testbuild:test testbuild:bak && docker stop mysite && docker rm mysite && docker run --name mysite -v site_logs:/var/log:rw -p 80:80 -d testbuild:test
]

Ansible

Ansible is more flexible option. Easy to read/diagnose.

using deployment/ansible/ for host docker provisionment

$ ansible-playbook mysite.yml

monitoring

Using an AWS health check to create an alarm if the instance is marked as "unhealth" /pics/profile_pic.png is the end point AWS polls on from multiple regions