should we do full blown nomad support? #180

Closed
opened 2025-02-15 03:36:45 +00:00 by despiegk · 0 comments
Owner

probably won't work I just see, license is not good https://github.com/hashicorp/nomad/blob/main/LICENSE

purpose

  • can we integrate mycelium as plugin to nomad
  • can we integrate flist as storage plugin to nomad
  • can we integrate the webgw as plugin
  • can we make some examples of how to use nomad
    • deploy postgresql cluster
    • deploy caddy
    • connect caddy to webgw
    • use flists as backend for the containers
    • internal communication between containers is mycelium
    • ...
  • can we use opentofu (or should it be other) to deploy required VM's on TFGrid which are then used in nomad?
  • or should we create a plugin to nomad which also exposes the VM creation so we don't need opentofu

requirements

  • super reliable, keeps on retrying to deploy VM's if it fails in TFGrid, is picky on selection on nodes
  • ability to chose another node if it would fail

questions

  • is license nomad ok?
  • next to our flist approach we need a distributed filesystem for access to e.g. LLM images which one is best and already supported, just found https://doc.beegfs.io/ but there are for sure more, other one https://moosefs.com/, seaweedfs (is a nice one)
  • can we create a public read only pool of info so its easy for people to start there AI llm/s
  • for the storage we also need usecase where it works over slower links
  • consults license is not good we can't use, can we use nomad without it?
> probably won't work I just see, license is not good https://github.com/hashicorp/nomad/blob/main/LICENSE ## purpose - can we integrate mycelium as plugin to nomad - can we integrate flist as storage plugin to nomad - can we integrate the webgw as plugin - can we make some examples of how to use nomad - deploy postgresql cluster - deploy caddy - connect caddy to webgw - use flists as backend for the containers - internal communication between containers is mycelium - ... - can we use opentofu (or should it be other) to deploy required VM's on TFGrid which are then used in nomad? - or should we create a plugin to nomad which also exposes the VM creation so we don't need opentofu ## requirements - [ ] super reliable, keeps on retrying to deploy VM's if it fails in TFGrid, is picky on selection on nodes - [ ] ability to chose another node if it would fail ## questions - is license nomad ok? - next to our flist approach we need a distributed filesystem for access to e.g. LLM images which one is best and already supported, just found https://doc.beegfs.io/ but there are for sure more, other one https://moosefs.com/, seaweedfs (is a nice one) - can we create a public read only pool of info so its easy for people to start there AI llm/s - for the storage we also need usecase where it works over slower links - consults license is not good we can't use, can we use nomad without it?
despiegk added this to the next milestone 2025-02-15 03:36:45 +00:00
despiegk changed title from full blown nomad support to should we do full blown nomad support? 2025-02-15 04:06:52 +00:00
Sign in to join this conversation.
No Milestone
No project
No Assignees
1 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: tfgrid/circle_engineering#180
No description provided.