Servers pool

/ infrastructure · compute
worker-03 at 94% 212 / 350 slots
Worker nodes
5
+ 1 brain node
App slots used
212 / 350
across the fleet
Fleet capacity
61%
fleet alert at 80%
Free slots
138
room for new apps

Fleet

large = 100 slots · standard = 50 Nomad healthy
ServerRole / sizeCPURAMApp slotsStatus
brain-01Nomad brain
— control only active
worker-01worker · large
48 / 10048%
active
worker-02worker · large
45 / 10045%
active
worker-03worker · standard
47 / 5094%
busy
worker-04worker · standard
30 / 5060%
draining
worker-05worker · standard
42 / 5084%
active

Each worker has an app-slot cap by size (large = 100, standard = 50). Placement bin-packs onto the emptiest node first; per-node alert at 90%, fleet alert at 80%. The fleet has 138 free slots, but worker-03 is nearly full — deploy fail-fast checks per-node capacity before building. Density is ultimately bounded by RAM & per-container limits, not just the count.

Add a server = clone the base image (Docker + Nomad pre-installed) → nomad node join → its slots join the pool. App data lives on a separate tier → see Databases.