NodeResourceDiagnostics
Resource readings for one node from gathered state and local psutil.
gatheredMemory object
Last event-sourced memory reading for this node.
- MemoryUsage
- null
ramTotal objectrequired
0ramAvailable objectrequired
0swapTotal objectrequired
0swapAvailable objectrequired
0currentMemory object
Live memory reading from the API process.
- MemoryUsage
- null
ramTotal objectrequired
0ramAvailable objectrequired
0swapTotal objectrequired
0swapAvailable objectrequired
0currentWired object
Live OS-level wired (unpageable) memory in use (macOS only). Read locally on this endpoint — deliberately NOT on the gossiped MemoryUsage, whose schema rides extra=forbid events — to detect leaked wired memory after an abnormal Metal termination (#239).
- Memory
- null
0disk object
Last event-sourced disk reading for this node.
- DiskUsage
- null
total objectrequired
0available objectrequired
0system object
Last event-sourced system performance reading.
- SystemPerformanceProfile
- null
00000accelerator object
- AcceleratorMetrics
- null
Possible values: [apple, amd, nvidia, intel, cpu, unknown]
unknownUnknownutilizationRatio object
- number
- null
vramTotalBytes object
- integer
- null
vramUsedBytes object
- integer
- null
gttTotalBytes object
GPU-mappable host (GTT) memory, for unified-memory APUs (e.g. AMD Strix
Halo). On such a node the GPU addresses system RAM beyond the BIOS VRAM
carve-out through GTT, so the usable GPU pool is far larger than
vram_total_bytes (placement uses this to admit big models on a UMA node).
None on discrete GPUs / collectors that do not report it.
- integer
- null
powerWatts object
- number
- null
temperatureCelsius object
- number
- null
clockMhz object
- integer
- null
computeCapability object
Discrete-GPU compute capability as "<major>.<minor>" (NVIDIA SM level),
e.g. "8.0" (A100 Ampere), "9.0" (H100 Hopper), "10.0" (B100/B200
Blackwell), "12.0" (RTX 50 Blackwell). The engine/quant/placement decision
keys on this, not on vendor: the same model+engine performs oppositely across
generations (a benchmark showed vLLM's MXFP4 path losing single-stream on
Ampere, which has no native FP4, but winning on Blackwell, which does).
None on collectors that do not report it (AMD sysfs, Apple).
- string
- null
nativeFp4 object
Whether the GPU accelerates FP4 natively (Blackwell sm100+, i.e. SM level
(major, minor) >= (10, 0)). Derived at the collector boundary from the parsed
compute capability (a numeric tuple compare, not a string compare). None
when unmeasured.
- boolean
- null
nativeFp8 object
Whether the GPU accelerates FP8 natively (Ada sm89 / Hopper sm90 and later).
Derived at the collector boundary from the compute capability. None when
unmeasured.
- boolean
- null
network object
Last event-sourced network interface reading.
- NodeNetworkInfo
- null
interfaces object[]
Possible values: [wifi, ethernet, maybe_ethernet, thunderbolt, unknown]
unknown{
"gatheredMemory": {
"ramTotal": {
"inBytes": 0
},
"ramAvailable": {
"inBytes": 0
},
"swapTotal": {
"inBytes": 0
},
"swapAvailable": {
"inBytes": 0
}
},
"currentMemory": {
"ramTotal": {
"inBytes": 0
},
"ramAvailable": {
"inBytes": 0
},
"swapTotal": {
"inBytes": 0
},
"swapAvailable": {
"inBytes": 0
}
},
"currentWired": {
"inBytes": 0
},
"disk": {
"total": {
"inBytes": 0
},
"available": {
"inBytes": 0
}
},
"system": {
"gpuUsage": 0,
"temp": 0,
"sysPower": 0,
"pcpuUsage": 0,
"ecpuUsage": 0,
"accelerator": {
"vendor": "unknown",
"name": "Unknown",
"utilizationRatio": 0,
"vramTotalBytes": 0,
"vramUsedBytes": 0,
"gttTotalBytes": 0,
"powerWatts": 0,
"temperatureCelsius": 0,
"clockMhz": 0,
"computeCapability": "string",
"nativeFp4": true,
"nativeFp8": true
}
},
"network": {
"interfaces": [
{
"name": "string",
"ipAddress": "string",
"interfaceType": "unknown"
}
]
}
}