@helhum

insight.helhum.io

@kaktusmimi

lihsmi.ch

The Big Picture

It's (all) about Feedback

• Establish and support a culture of feedback in your development teams
• Enable pair-programming sessions
• Establish a code-review workflow

Small Batches

• Work in small batches
• Smaller batches result in faster feedback
• Big changes make review a cumbersome thing to do
• [SMELL] Long living feature branches

Cross-reference Issues

• Integrate your development workflow with the project management workflow
• Label your commits with your ticket numbers
• Link code changes with tickets and vice versa
• Label (acceptance) tests with ticket numbers

Version Control Workflow

• There are many workflows
• It doesn't matter which one you chose
• Chose one and stick to it

Recommendation Gitlab Merge Request Workflow

• Easy to implement
• Fully supported by open source Gitlab
• Supports short living feature branches
• Little to no learning required

Gitlab Merge Requests

• Create a feature branch (PRJ-123)
• Implement your feature
• Push branch to Gitlab server
• Create merge request
• Assign it to your colleague
• Merge or decline

Keep everything in Version Control

• Configuration & testdata
• Continuous Integration Jobs
• Infrastructure Code
• Deployment Scripts
• [SMELL] Configuration in database (TS)

"Environmental Awareness"

• Development
• Integration
• Staging
• Production

Development Environment

• Maximize log output
• Disable all caching
• Provide production-like environment
  • OS, Packages, Versions...

Testing Environment

• Use separate database
  • Don't "override" development data
• Use production context (Flow)
  • Might make things a lot faster

Staging Environment

• For customer review
• Stick to production settings
• Remember access restrictions
• robots.txt might be a good idea!

Production

• Turn off logging
• Enable caches

Implementing Environments

• Use TYPO3_CONTEXT
• Use environment variables
  • Apache set_env TYPO3_CONTEXT
  • CLI export TYPO3_CONTEXT
• Include different Settings files

Prepare for Update and Disaster Recovery

• Use Configuration Management if possible
• At least: Document setup very well

Build Tasks

• Composer
• CSS, JavaScript, Sprites
• Database migrations
• Database import / export

Testing Principles

• Avoid manual acceptance testing
• Take care of your testdata
• Write tests
  • isolated
  • independent of execution order

Unit Tests

• Build the thing right?
• Offer very specific feedback
• Are very fast

End-to-End Tests

• Build the right thing?
• Tell you whether your application works
• Are quite slow

Testing doesn't stop with development

• Monitoring tests
  • Is your application alive?
• Performance tests
  • How fast is my website
  • Did the last deployment break something?

How to handle User Data

• Independent directory structure for user data
• Separate project files from user data
  • Templates
  • CSS
  • Sprites, Icons...

Where to build stuff

• Don't build things on (production) servers
• Use integration server
• Ship as package
• [SMELL] git on (production) servers

Rollback

• Rollback for files is easy
• Rollback for database is hard
• Prepare for disaster recovery

Handling of external services

• Solr needs configuration & restart
• Cache clearing in
  • Varnish
  • Memcached / Redis

Recommended Tool

• TYPO3 Surf
• Many best-practices and pre-defined tools for
  • TYPO3
  • Neos
  • Flow

Profiling vs. Monitoring

• Profiling is analyzing single requests
  • Mostly during development
• Monitoring is an aggregated view on your application
  • Mostly during operations

Monitoring Task Examples

• Is the homepage showing
• Does the FE user login work
• Does the search work
• Is the tt_news ticker showing some news
• Are there any severe log entries

How to implement
Monitoring Tasks

• Use rock solid technology
  • Selenium is fragile!
• Use Jenkins as a task runner
• Send emails or IM messages to developers
• Set up dashboards

Third-party Applications

• gtmetrix for page speed
• NewRelic for full stack monitoring / profiling
• Gatling for performance testing

Logfile management

• Log files provide a valuable source of information
• Often scattered all over the system
• Use logfile management tools like
  • Logstash / Heka
  • ElasticSearch
  • Kibana

Books

Books

Books