1.6 KiB
Changes in the workforce!
- Taeer is leaving modus / fediversity
- Nicolas is joining in his stead
- Hans is also coming back
- Jimena is leaving the project as well.
Architecture document
- "Different design choices" A.K.A. shortcuts
-
Before the idea was that all you needed was a pile of hardware, and our application would take care of everything on top of that with NixOS
-
Now: use a layer in between for virtualisation, clustering, etc. that's already been invented
-
Kubernetes: too complicated
-
proxmox: not as complicated
- Before: share services between organizations
- Now: each organization has its own set of services
Question: How do you handle inter-service communication? Example: Matrix, authentication mechanism. Do they run together or are they separate services? DNS (TBD), Email (stalwart), Authentication (TBD), Backend storage (garage).
Bogdan: unclear about the implications of using Proxmox
Original question regarding kubernetes/nixops was: can we design around features that kubernetes has
Is the feature of kubernetes of dynamic scaling quickly is actually something that is currently needed in fediversity architecture proxmos allows scaling in virtual machines, (missed a part here), and is already familiar for lots of people
it also gives api to deploy that nixops can communicate with
Koen will adjust the architecture document next week to include the Proxmos, and then meet with Taeer and Nicolas on Wednesday Sept 4th at 4pm NL time
Communication Koen wants daily updates posted publicly, but this is still open for discussion Posting updates to our existing fediverse instances