- Zebra
- Per Printhouse situation
- Cloud Situation
- Disaster Recovery
- Automation
- Future of zebra?
- Cloud rendering?
- Luigi
- Who is this guy and why should I care?
- Platform arch. on AWS (VPCs / ELBs / SGs / ASGs / LCs )
- Services
- Golumn (Gateway / PBs / Stitcher )
- Java (Generic Renderer)
- C++ (Dynamic Renderers)
- Service Discovery (Consul)
- Tooling
- Lambda deregistration
- AWS Cost Calculator (+Periscope)
- Kapacitor :(
- Services
- Ansible
- Setup + workflow
- Chef
- Overview of structure
- Workflow
- Old platform arch (will die during Q4 hopefully)
- Tars
- Solar
- Landmarks
- Workflows (from dev workstation to prod)
- Workflow for chef code
- Workflow for platform code (DM)
- Workflow for legacy projects code? (RPM)
- CI systems
- Travis
- Jenkins
- Jenkins II
- Deployment Manager
- Overview of what it is and what it does
- How is hooked up to our deployment pipeline?
- Travis/Jenkins to S3 + DM
- Nodes registering
- DM notifying nodes (chefstarter)
- Readyscale Setup
- ELK
- Zabbix
- Grafana + Influx
- Pager Duty
- Sadly, 95% of the above
- There's part of zebra already documented, but they need updating and cleaning up
- Ansible + Chef: They are infrastructure as code, so while not having a documentation, it's easy to get up and running while reading the code
- Readyscale setup, no docs, no code, no nothing... just my understanding of what's in there by diving into it... I'll rather not document this, but propose a plan for getting rid of it (move to a chef+ansible managed solution inside our main AWS account. Mainly, migrate zabbix/ELK and destroy the rest)
- Better monitoring + Smarter alerting
- Zabbix integrated better with AWS/Platform
- Intelligent alerting on metrics (Influx Kapacitor)
- Disaster Recovery
- Simian Army!
- Schedule 1 every 2 weeks a planned downtime on staging, and work with the team on recovering the service...
- Cost Savings
- Containerization?
- Multiple services x box
- Smoothly spin up/down envs
- ...??
- Separate asset management from Rendering
- Landmarks DB -> how to make it HA if it's not in the rendering env?
(Grant him access to everything)
- Organize sessions from Things to be handed over during the initial 2 weeks over the next topics:
- Overview of our architecture (Infrastructure wise) 2,3,6,10
- Overview of our architecture (Platform wise) (+ Rob || Tony || Cms || Jelmer)
- Go through our workflows for pushing infrastructure code (Ansible + Chef) 4,5,8
- Go through our workflows for pushing application code (GO/Java/C++ + DM) 7,8,9
- Go over our grafana/kibana dashboards 10,11
- Long session to over Zebra 1
- OnCall 12
- Schedule regular sessions during the first 2 week span, to allow him to have a forum where to freely ask about anything (midday?) (apart of obviously support him during his day to day)
- 2 sessions per week, 90m per session.
- Follow up the sessions with small tasks that will allow him to get hands on the topics discussed.
- Sessions with the rest of the teams to understand the rest of the company (30-45m max).
- 1 open session everyday (13:30 - 14:00) - Ask any questions and track progress
week | infra meetings | team meetings | tasks |
---|---|---|---|
29 Sept | Structure Overview | TBD | |
1 | Ansible + Chef | TBD | |
2 | Alerting/OnCall + Metrics/Logging | QA | TBD |
3 | Zebra + PSP | Product Delivery | TBD |
4 | --- | Shop + BI | TBD |
- DM: Add which version is deployed on the dropdown when choosing a branch-commit
- Chef: Damn reload bug?