Systems
Reading live sources…
Systems
Reading live sources…
Reading live data…
The K3s nodes that FlexInfer, the Mills factory, and this site run on: what each node offers and how the pods are doing. Read from FlexDeck when the page loads, with ids and names anonymised.
15 nodes, 15 ready, with 156 cores and 418 GiB of memory between them. Cluster age: 2+ years. Capacity is what each node offers to Kubernetes, not what is in use.
Hover, focus, or tap a node to see where its sampled pods sit. FlexDeck publishes 200 of the cluster’s 627 pods as a sample balanced across categories, so these counts show placement, not load.
| Node | Role | Status | CPU | Memory | Sampled pods | By category |
|---|---|---|---|---|---|---|
| control-plane-4Aj6 | Control plane | Ready | 4 cores | 7.8 GiB | 5 (5 running) | 2 monitoring, 3 infra |
| gpu-worker-wo_6 | GPU workers | Ready | 28 cores | 126 GiB | 19 (18 running) | 1 ai, 12 app, 3 monitoring, 3 infra |
| gpu-worker-Qg_Z | GPU workers | Ready | 28 cores | 62.7 GiB | 9 (9 running) | 1 ai, 1 app, 3 monitoring, 4 infra |
| gpu-worker-AoLZ | GPU workers | Ready | 24 cores | 62.0 GiB | 11 (11 running) | 1 ai, 6 app, 2 monitoring, 2 infra |
| gpu-worker-6x1f | GPU workers | Ready | 8 cores | 15.6 GiB | 6 (6 running) | 1 ai, 3 monitoring, 2 infra |
| worker-f8o2 | Workers | Ready | 8 cores | 15.6 GiB | 33 (33 running) | 10 ai, 6 app, 5 monitoring, 12 infra |
| worker-FaR1 | Workers | Ready | 8 cores | 15.6 GiB | 9 (9 running) | 5 app, 3 monitoring, 1 infra |
| worker-xICt | Workers | Ready | 8 cores | 15.6 GiB | 7 (7 running) | 3 app, 3 monitoring, 1 infra |
| worker-hMth | Workers | Ready | 8 cores | 15.6 GiB | 24 (9 running) | 1 ai, 14 app, 8 monitoring, 1 infra |
| worker-6MOa | Workers | Ready | 8 cores | 13.6 GiB | 6 (6 running) | 1 app, 3 monitoring, 2 infra |
| worker-Le33 | Workers | Ready | 8 cores | 13.6 GiB | 5 (5 running) | 1 app, 2 monitoring, 2 infra |
| worker-PhO7 | Workers | Ready | 4 cores | 15.6 GiB | 18 (17 running) | 2 ai, 1 app, 6 monitoring, 9 infra |
| worker-tWlm | Workers | Ready | 4 cores | 15.6 GiB | 20 (20 running) | 4 ai, 3 app, 7 monitoring, 6 infra |
| worker-LQd5 | Workers | Ready | 4 cores | 11.7 GiB | 14 (14 running) | 3 ai, 4 app, 4 monitoring, 3 infra |
| worker-rELl | Workers | Ready | 4 cores | 11.7 GiB | 14 (14 running) | 3 ai, 1 app, 4 monitoring, 6 infra |
98 of 100 is the share of the cluster’s 627 pods in the Running phase. Finished job pods are left out of the denominator, so scheduled jobs do not pull a healthy cluster below 100.
The node list above is current. The physical layer under the nodes is not read live: the last map is a committed snapshot from . At that point the cluster had 16 nodes: 4 physical machines, including the GPU nodes (2× AMD 7900 XTX, AMD Radeon VII and NVIDIA GTX 980 Ti), and 12 virtual machines on one Harvester host (a Dell R730xd with 256 GB of memory). Read it as the shape of the hardware, not today’s inventory.
This page renders the same documents the public API serves, after the anonymizers in lib/public-api/anonymize.ts have run on the server.
The models running on these nodes are on the inference plane page.
© 2026 FlexInfer.