# agenda for the 1st libre.sh call > 9th April 1PM UTC [TOC] ## attendance - Pierre (indiehosters) - migrating to beautiful Kubernetes, working with it since 1a as a freelancer - mattronix - Matthew (wehost) - Keycloak, LDAP authentication (using openldap and keycloak in production), networking, k8s, data management, IT Process Management, Linux, Ansible, Documentation, “cloud” native architecture - realitygaps (wehost) - libre hosting, free software, ~~best~~ good practices, sharing knowledge between groups, development/ci - yala / almereyda - Jon (Ecobytes) - libre.sh user and interested in long-term maintenance - Samuel (Lalop) - curiosity, developer - gandhiano - Gualter - ~ 10a with Ecobytes, working with Ansible and K8S professionally, interest in building a K8S cluster for librehosters - MH (Molus) - developer, analyst, building a FLOSS version of "Google Cloud" ## IndieHosters history - 4 yrs ago on Hetzner ## technical ### libre.sh 1.0 docker compose - LTS by whom? - @yala - backup format of 1 could bring a nice migration path towards 2, once restoration is implemented ### libre.sh 2.0 kubernetes #### Use Case difficult for people who don't want to host on public clouds > How to build declarative infrastructure who don't want to host on public clouds? #### Issues - encryption of the network was not given until recently for `flannel` - how to mount storage into pods/containers? (stateful applications) - rook.io is not alpha anymore, now beta a way to make Ceph K8S native - a way to have our own cluster on bare metal machines #### Components being worked on - 2 epyc servers from hetzner for free! - contact with David from `hetzner-kube` student project - upstream Pull Requests - cert-manager - rook - discourse #### Questions :::warning See below for organisational points. ::: - How to upgrade? and why to upgrade? kube-charts vs libresh-v1 - Where to deploy K8S? hetzner vs own infrastructure vs raspberries@home vs public cloud? - bare metal? - what about pxe and old servers? - how to lay out different machine specifics? - nodes for high-speed database services - nodes for different constaints - nodes for storage - heterogenous nodes? - federation of clusters? - public clouds AWS, Azure, Google Cloud, Digitalocean, ... - different maintainer communities - libre.sh will be ... - a set of good practices to deploy k8s without any cloud provider dependencies - a K8S distribution - logging - monitoring - persistent block and object storage - easy Backup and Recovery for export of data aiming at the old idea of :bulb: migration networks - Single Sign-on - a collection of application recipies #### open questions - [ ] shared `staging` cluster - everybody has read/write access - public services, with limited warranty - https://github.com/libresh/libresh-staging/blob/master/README.md - [ ] shared `production` cluster - ecobytes - indie - wehost own hardware in a cooperative data centre - ? - [ ] GitHub vs. GitLab - git trees - issue tracker - Kanban - technical in GitHub projects - organisational in WeKan - (CI/CD) - [ ] CI/CD - gitlab-ci - jenkins - spinaker? interesting as its purpose built - [ ] last doubt, where to store application data - git - K8S API - our own API? - Backup and Recovery? - Docker Registry management and distrubtion - [ ] Security Policy - role-based access control - [ ] Privacy Policy ## organisational R&D Process? (Track based development based on component development) Project Management (Track Overview and coordination) Technical Team and assignemt (work distrubition skillset inventory and utilization of team members) Training and Content (training videos, documentation) ### Informational and infrastructure design - Identity - Manageability - Naming Conventions, Standardization - Data Storage locations (structure, prefix) - Silos (Database, Applications, Storage) - Patching and Bug Fixing ### Social organisation - easy adoption and management of services - self service interface for app deployments on K8S? i.e. Rancher - targeting the tech savvy linked to identity management there is also a market here - Project Process and Structure - Team Structure (data team content management and persistence), (application management, application certification and distrubution), (custom development, Selfservice and services) ### Communication - timeline for the "press release" Blog posts on third-party blogs and low-tech Markdown somewhere - how to install libre.sh on hetzner 2 epyc servers requested and donated by them - `rook` performance developer contacted and support available - Nextcloud performance on top of `rook` contacted, no reply yet, wehost have contacts - Catalogue maybe later? @maxlath informed about unavailability for contractual work, but seconded the availability for quick bugfixing in `wikidata-sdk` - open collective page - logo designing a logo and banner, setting up the tiers, writing updates, providing an informative about text - explanation why this is useful to the uninformed public a federated workflow engine and reactive infrastructure for ad-hoc and long-term civic cooperation bootstrapping output of the *Librehosting syndicate* (in foundational declaration phase), which can only be proven after the fact - Mastodon? - who is interested in working on what - media: @molus (writing release notes) - Hacker News, Reddit - after each meeting release what we discussed - k8s: @gandhiano, @pierre, @molus - bridge Matrix channels to Freenode ### Resources and costs machines, time - donated by the participating collectives # possible dates/time Pierre - weekdays - 8AM-17:30PM UTC. Jon - From March 26 on daily from 13 to 17 CEST MH (Molus) - available 1PM to 10PM UTC Samuel (Lalop) - available from 8:30AM to 4PM UTC realitygaps - monday or friday afternoon Matthew Frost (mattronix) - monday or friday afternoon gandhiano - random availability (very flexible on children-free weeks, very difficult on childcare weeks) Suggested dates: 9-12 April 2018 # minutes of meeting Volunteer requested to take minutes # Scheduling next steps Next call is starting a rhythm of regular check-ins for the community. - bi-weekly - more actionable points, hands-on - experiment and get confident - putting workload - trying some failure modes - decision making process means to - have fun first! - act - allow to object by inviting regularily to review rounds two next meetings - [ ] technical development https://hack.allmende.io/libre.sh-20180423-technical-community-check-in - [ ] organisational maintenance https://hack.allmende.io/libre.sh-201804-organisational-check-in @mattronix, @yala