mirror of
https://github.com/meta-llama/llama-stack.git
synced 2025-12-03 09:53:45 +00:00
Merge branch 'main' into routeur
This commit is contained in:
Sébastien Han
GitHub
commit
3ce509e94a
87 changed files with 67526 additions and 4478 deletions
|
|
@ -171,10 +171,18 @@ if [[ "$COLLECT_ONLY" == false ]]; then
|
|||
|
||||
# Set MCP host for in-process MCP server tests
|
||||
# - For library client and server mode: localhost (both on same host)
|
||||
# - For docker mode: host.docker.internal (container needs to reach host)
|
||||
# - For docker mode on Linux: localhost (container uses host network, shares network namespace)
|
||||
# - For docker mode on macOS/Windows: host.docker.internal (container uses bridge network)
|
||||
if [[ "$STACK_CONFIG" == docker:* ]]; then
|
||||
export LLAMA_STACK_TEST_MCP_HOST="host.docker.internal"
|
||||
echo "Setting MCP host: host.docker.internal (docker mode)"
|
||||
if [[ "$(uname)" != "Darwin" ]] && [[ "$(uname)" != *"MINGW"* ]]; then
|
||||
# On Linux with host network mode, container shares host network namespace
|
||||
export LLAMA_STACK_TEST_MCP_HOST="localhost"
|
||||
echo "Setting MCP host: localhost (docker mode with host network)"
|
||||
else
|
||||
# On macOS/Windows with bridge network, need special host access
|
||||
export LLAMA_STACK_TEST_MCP_HOST="host.docker.internal"
|
||||
echo "Setting MCP host: host.docker.internal (docker mode with bridge network)"
|
||||
fi
|
||||
else
|
||||
export LLAMA_STACK_TEST_MCP_HOST="localhost"
|
||||
echo "Setting MCP host: localhost (library/server mode)"
|
||||
|
|
|
|||
|
|
@ -8,7 +8,6 @@
|
|||
Schema discovery and collection for OpenAPI generation.
|
||||
"""
|
||||
|
||||
import importlib
|
||||
from typing import Any
|
||||
|
||||
|
||||
|
|
@ -20,23 +19,6 @@ def _ensure_components_schemas(openapi_schema: dict[str, Any]) -> None:
|
|||
openapi_schema["components"]["schemas"] = {}
|
||||
|
||||
|
||||
def _load_extra_schema_modules() -> None:
|
||||
"""
|
||||
Import modules outside llama_stack_api that use schema_utils to register schemas.
|
||||
|
||||
The API package already imports its submodules via __init__, but server-side modules
|
||||
like telemetry need to be imported explicitly so their decorator side effects run.
|
||||
"""
|
||||
extra_modules = [
|
||||
"llama_stack.core.telemetry.telemetry",
|
||||
]
|
||||
for module_name in extra_modules:
|
||||
try:
|
||||
importlib.import_module(module_name)
|
||||
except ImportError:
|
||||
continue
|
||||
|
||||
|
||||
def _extract_and_fix_defs(schema: dict[str, Any], openapi_schema: dict[str, Any]) -> None:
|
||||
"""
|
||||
Extract $defs from a schema, move them to components/schemas, and fix references.
|
||||
|
|
@ -79,9 +61,6 @@ def _ensure_json_schema_types_included(openapi_schema: dict[str, Any]) -> dict[s
|
|||
iter_registered_schema_types,
|
||||
)
|
||||
|
||||
# Import extra modules (e.g., telemetry) whose schema registrations live outside llama_stack_api
|
||||
_load_extra_schema_modules()
|
||||
|
||||
# Handle explicitly registered schemas first (union types, Annotated structs, etc.)
|
||||
for registration_info in iter_registered_schema_types():
|
||||
schema_type = registration_info.type
|
||||
|
|
|
|||
|
|
@ -1,11 +1,24 @@
|
|||
{
|
||||
"annotations": {
|
||||
"list": []
|
||||
"list": [
|
||||
{
|
||||
"builtIn": 1,
|
||||
"datasource": {
|
||||
"type": "grafana",
|
||||
"uid": "-- Grafana --"
|
||||
},
|
||||
"enable": true,
|
||||
"hide": true,
|
||||
"iconColor": "rgba(0, 211, 255, 1)",
|
||||
"name": "Annotations & Alerts",
|
||||
"type": "dashboard"
|
||||
}
|
||||
]
|
||||
},
|
||||
"editable": true,
|
||||
"fiscalYearStartMonth": 0,
|
||||
"graphTooltip": 0,
|
||||
"id": null,
|
||||
"id": 1,
|
||||
"links": [],
|
||||
"liveNow": false,
|
||||
"panels": [
|
||||
|
|
@ -16,11 +29,40 @@
|
|||
},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {
|
||||
"mode": "palette-classic"
|
||||
},
|
||||
"custom": {
|
||||
"axisBorderShow": false,
|
||||
"axisCenteredZero": false,
|
||||
"axisColorMode": "text",
|
||||
"axisLabel": "",
|
||||
"axisPlacement": "auto",
|
||||
"barAlignment": 0,
|
||||
"drawStyle": "line",
|
||||
"fillOpacity": 10,
|
||||
"gradientMode": "none",
|
||||
"hideFrom": {
|
||||
"legend": false,
|
||||
"tooltip": false,
|
||||
"viz": false
|
||||
},
|
||||
"insertNulls": false,
|
||||
"lineInterpolation": "linear",
|
||||
"lineWidth": 1,
|
||||
"pointSize": 5,
|
||||
"scaleDistribution": {
|
||||
"type": "linear"
|
||||
},
|
||||
"showPoints": "auto",
|
||||
"fillOpacity": 10
|
||||
"spanNulls": false,
|
||||
"stacking": {
|
||||
"group": "A",
|
||||
"mode": "none"
|
||||
},
|
||||
"thresholdsStyle": {
|
||||
"mode": "off"
|
||||
}
|
||||
},
|
||||
"mappings": [],
|
||||
"thresholds": {
|
||||
|
|
@ -32,7 +74,8 @@
|
|||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {
|
||||
"h": 8,
|
||||
|
|
@ -40,15 +83,16 @@
|
|||
"x": 0,
|
||||
"y": 0
|
||||
},
|
||||
"id": 1,
|
||||
"id": 2,
|
||||
"options": {
|
||||
"legend": {
|
||||
"calcs": [],
|
||||
"displayMode": "table",
|
||||
"displayMode": "list",
|
||||
"placement": "bottom",
|
||||
"showLegend": true
|
||||
},
|
||||
"tooltip": {
|
||||
"maxHeight": 600,
|
||||
"mode": "multi",
|
||||
"sort": "none"
|
||||
}
|
||||
|
|
@ -59,9 +103,112 @@
|
|||
"type": "prometheus",
|
||||
"uid": "prometheus"
|
||||
},
|
||||
"expr": "llama_stack_completion_tokens_total",
|
||||
"legendFormat": "{{model_id}} ({{provider_id}})",
|
||||
"refId": "A"
|
||||
"disableTextWrap": false,
|
||||
"editorMode": "builder",
|
||||
"expr": "sum by(gen_ai_request_model) (llama_stack_gen_ai_client_token_usage_sum{gen_ai_token_type=\"input\"})",
|
||||
"fullMetaSearch": false,
|
||||
"includeNullMetadata": true,
|
||||
"legendFormat": "__auto",
|
||||
"range": true,
|
||||
"refId": "A",
|
||||
"useBackend": false
|
||||
}
|
||||
],
|
||||
"title": "Prompt Tokens",
|
||||
"type": "timeseries"
|
||||
},
|
||||
{
|
||||
"datasource": {
|
||||
"type": "prometheus",
|
||||
"uid": "prometheus"
|
||||
},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {
|
||||
"mode": "palette-classic"
|
||||
},
|
||||
"custom": {
|
||||
"axisBorderShow": false,
|
||||
"axisCenteredZero": false,
|
||||
"axisColorMode": "text",
|
||||
"axisLabel": "",
|
||||
"axisPlacement": "auto",
|
||||
"barAlignment": 0,
|
||||
"drawStyle": "line",
|
||||
"fillOpacity": 10,
|
||||
"gradientMode": "none",
|
||||
"hideFrom": {
|
||||
"legend": false,
|
||||
"tooltip": false,
|
||||
"viz": false
|
||||
},
|
||||
"insertNulls": false,
|
||||
"lineInterpolation": "linear",
|
||||
"lineWidth": 1,
|
||||
"pointSize": 5,
|
||||
"scaleDistribution": {
|
||||
"type": "linear"
|
||||
},
|
||||
"showPoints": "auto",
|
||||
"spanNulls": false,
|
||||
"stacking": {
|
||||
"group": "A",
|
||||
"mode": "none"
|
||||
},
|
||||
"thresholdsStyle": {
|
||||
"mode": "off"
|
||||
}
|
||||
},
|
||||
"mappings": [],
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{
|
||||
"color": "green",
|
||||
"value": null
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {
|
||||
"h": 8,
|
||||
"w": 12,
|
||||
"x": 12,
|
||||
"y": 0
|
||||
},
|
||||
"id": 1,
|
||||
"options": {
|
||||
"legend": {
|
||||
"calcs": [],
|
||||
"displayMode": "list",
|
||||
"placement": "bottom",
|
||||
"showLegend": true
|
||||
},
|
||||
"tooltip": {
|
||||
"maxHeight": 600,
|
||||
"mode": "multi",
|
||||
"sort": "none"
|
||||
}
|
||||
},
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {
|
||||
"type": "prometheus",
|
||||
"uid": "prometheus"
|
||||
},
|
||||
"disableTextWrap": false,
|
||||
"editorMode": "builder",
|
||||
"exemplar": false,
|
||||
"expr": "sum by(gen_ai_request_model) (llama_stack_gen_ai_client_token_usage_sum{gen_ai_token_type=\"output\"})",
|
||||
"fullMetaSearch": false,
|
||||
"includeNullMetadata": true,
|
||||
"interval": "",
|
||||
"legendFormat": "__auto",
|
||||
"range": true,
|
||||
"refId": "A",
|
||||
"useBackend": false
|
||||
}
|
||||
],
|
||||
"title": "Completion Tokens",
|
||||
|
|
@ -74,78 +221,40 @@
|
|||
},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {
|
||||
"mode": "palette-classic"
|
||||
},
|
||||
"custom": {
|
||||
"axisBorderShow": false,
|
||||
"axisCenteredZero": false,
|
||||
"axisColorMode": "text",
|
||||
"axisLabel": "",
|
||||
"axisPlacement": "auto",
|
||||
"barAlignment": 0,
|
||||
"drawStyle": "line",
|
||||
"fillOpacity": 10,
|
||||
"gradientMode": "none",
|
||||
"hideFrom": {
|
||||
"legend": false,
|
||||
"tooltip": false,
|
||||
"viz": false
|
||||
},
|
||||
"insertNulls": false,
|
||||
"lineInterpolation": "linear",
|
||||
"lineWidth": 1,
|
||||
"pointSize": 5,
|
||||
"scaleDistribution": {
|
||||
"type": "linear"
|
||||
},
|
||||
"showPoints": "auto",
|
||||
"fillOpacity": 10
|
||||
},
|
||||
"mappings": [],
|
||||
"thresholds": {
|
||||
"mode": "absolute",
|
||||
"steps": [
|
||||
{
|
||||
"color": "green",
|
||||
"value": null
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
"gridPos": {
|
||||
"h": 8,
|
||||
"w": 12,
|
||||
"x": 12,
|
||||
"y": 0
|
||||
},
|
||||
"id": 2,
|
||||
"options": {
|
||||
"legend": {
|
||||
"calcs": [],
|
||||
"displayMode": "table",
|
||||
"placement": "bottom",
|
||||
"showLegend": true
|
||||
},
|
||||
"tooltip": {
|
||||
"mode": "multi",
|
||||
"sort": "none"
|
||||
}
|
||||
},
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {
|
||||
"type": "prometheus",
|
||||
"uid": "prometheus"
|
||||
},
|
||||
"expr": "llama_stack_prompt_tokens_total",
|
||||
"legendFormat": "Prompt - {{model_id}}",
|
||||
"refId": "A"
|
||||
},
|
||||
{
|
||||
"datasource": {
|
||||
"type": "prometheus",
|
||||
"uid": "prometheus"
|
||||
},
|
||||
"expr": "llama_stack_tokens_total",
|
||||
"legendFormat": "Total - {{model_id}}",
|
||||
"refId": "B"
|
||||
}
|
||||
],
|
||||
"title": "Prompt & Total Tokens",
|
||||
"type": "timeseries"
|
||||
},
|
||||
{
|
||||
"datasource": {
|
||||
"type": "prometheus",
|
||||
"uid": "prometheus"
|
||||
},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"custom": {
|
||||
"drawStyle": "line",
|
||||
"lineInterpolation": "linear",
|
||||
"showPoints": "auto",
|
||||
"fillOpacity": 10
|
||||
"spanNulls": false,
|
||||
"stacking": {
|
||||
"group": "A",
|
||||
"mode": "none"
|
||||
},
|
||||
"thresholdsStyle": {
|
||||
"mode": "off"
|
||||
}
|
||||
},
|
||||
"mappings": [],
|
||||
"thresholds": {
|
||||
|
|
@ -158,7 +267,8 @@
|
|||
]
|
||||
},
|
||||
"unit": "ms"
|
||||
}
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {
|
||||
"h": 8,
|
||||
|
|
@ -175,6 +285,7 @@
|
|||
"showLegend": true
|
||||
},
|
||||
"tooltip": {
|
||||
"maxHeight": 600,
|
||||
"mode": "multi",
|
||||
"sort": "none"
|
||||
}
|
||||
|
|
@ -219,7 +330,8 @@
|
|||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {
|
||||
"h": 8,
|
||||
|
|
@ -240,8 +352,11 @@
|
|||
"fields": "",
|
||||
"values": false
|
||||
},
|
||||
"textMode": "auto"
|
||||
"showPercentChange": false,
|
||||
"textMode": "auto",
|
||||
"wideLayout": true
|
||||
},
|
||||
"pluginVersion": "11.0.0",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {
|
||||
|
|
@ -272,7 +387,8 @@
|
|||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {
|
||||
"h": 8,
|
||||
|
|
@ -293,8 +409,11 @@
|
|||
"fields": "",
|
||||
"values": false
|
||||
},
|
||||
"textMode": "auto"
|
||||
"showPercentChange": false,
|
||||
"textMode": "auto",
|
||||
"wideLayout": true
|
||||
},
|
||||
"pluginVersion": "11.0.0",
|
||||
"targets": [
|
||||
{
|
||||
"datasource": {
|
||||
|
|
@ -315,11 +434,40 @@
|
|||
},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {
|
||||
"mode": "palette-classic"
|
||||
},
|
||||
"custom": {
|
||||
"axisBorderShow": false,
|
||||
"axisCenteredZero": false,
|
||||
"axisColorMode": "text",
|
||||
"axisLabel": "",
|
||||
"axisPlacement": "auto",
|
||||
"barAlignment": 0,
|
||||
"drawStyle": "line",
|
||||
"fillOpacity": 10,
|
||||
"gradientMode": "none",
|
||||
"hideFrom": {
|
||||
"legend": false,
|
||||
"tooltip": false,
|
||||
"viz": false
|
||||
},
|
||||
"insertNulls": false,
|
||||
"lineInterpolation": "linear",
|
||||
"lineWidth": 1,
|
||||
"pointSize": 5,
|
||||
"scaleDistribution": {
|
||||
"type": "linear"
|
||||
},
|
||||
"showPoints": "auto",
|
||||
"fillOpacity": 10
|
||||
"spanNulls": false,
|
||||
"stacking": {
|
||||
"group": "A",
|
||||
"mode": "none"
|
||||
},
|
||||
"thresholdsStyle": {
|
||||
"mode": "off"
|
||||
}
|
||||
},
|
||||
"mappings": [],
|
||||
"thresholds": {
|
||||
|
|
@ -332,7 +480,8 @@
|
|||
]
|
||||
},
|
||||
"unit": "reqps"
|
||||
}
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {
|
||||
"h": 8,
|
||||
|
|
@ -349,6 +498,7 @@
|
|||
"showLegend": true
|
||||
},
|
||||
"tooltip": {
|
||||
"maxHeight": 600,
|
||||
"mode": "multi",
|
||||
"sort": "none"
|
||||
}
|
||||
|
|
@ -374,11 +524,40 @@
|
|||
},
|
||||
"fieldConfig": {
|
||||
"defaults": {
|
||||
"color": {
|
||||
"mode": "palette-classic"
|
||||
},
|
||||
"custom": {
|
||||
"axisBorderShow": false,
|
||||
"axisCenteredZero": false,
|
||||
"axisColorMode": "text",
|
||||
"axisLabel": "",
|
||||
"axisPlacement": "auto",
|
||||
"barAlignment": 0,
|
||||
"drawStyle": "line",
|
||||
"fillOpacity": 10,
|
||||
"gradientMode": "none",
|
||||
"hideFrom": {
|
||||
"legend": false,
|
||||
"tooltip": false,
|
||||
"viz": false
|
||||
},
|
||||
"insertNulls": false,
|
||||
"lineInterpolation": "linear",
|
||||
"lineWidth": 1,
|
||||
"pointSize": 5,
|
||||
"scaleDistribution": {
|
||||
"type": "linear"
|
||||
},
|
||||
"showPoints": "auto",
|
||||
"fillOpacity": 10
|
||||
"spanNulls": false,
|
||||
"stacking": {
|
||||
"group": "A",
|
||||
"mode": "none"
|
||||
},
|
||||
"thresholdsStyle": {
|
||||
"mode": "off"
|
||||
}
|
||||
},
|
||||
"mappings": [],
|
||||
"thresholds": {
|
||||
|
|
@ -391,7 +570,8 @@
|
|||
]
|
||||
},
|
||||
"unit": "Bps"
|
||||
}
|
||||
},
|
||||
"overrides": []
|
||||
},
|
||||
"gridPos": {
|
||||
"h": 8,
|
||||
|
|
@ -408,6 +588,7 @@
|
|||
"showLegend": true
|
||||
},
|
||||
"tooltip": {
|
||||
"maxHeight": 600,
|
||||
"mode": "multi",
|
||||
"sort": "none"
|
||||
}
|
||||
|
|
@ -437,7 +618,7 @@
|
|||
}
|
||||
],
|
||||
"refresh": "5s",
|
||||
"schemaVersion": 38,
|
||||
"schemaVersion": 39,
|
||||
"tags": [
|
||||
"llama-stack"
|
||||
],
|
||||
|
|
@ -445,13 +626,14 @@
|
|||
"list": []
|
||||
},
|
||||
"time": {
|
||||
"from": "now-15m",
|
||||
"from": "now-3h",
|
||||
"to": "now"
|
||||
},
|
||||
"timeRangeUpdatedDuringEditOrView": false,
|
||||
"timepicker": {},
|
||||
"timezone": "browser",
|
||||
"title": "Llama Stack Metrics",
|
||||
"uid": "llama-stack-metrics",
|
||||
"version": 0,
|
||||
"version": 17,
|
||||
"weekStart": ""
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue