mirror of
https://github.com/katanemo/plano.git
synced 2026-04-30 03:16:28 +02:00
support session pinning for consistent model selection in routing (#813)
This commit is contained in:
parent
785bf7e021
commit
46a5bfd82d
8 changed files with 406 additions and 3 deletions
|
|
@ -55,6 +55,63 @@ Response:
|
|||
|
||||
The response tells you which model would handle this request and which route was matched, without actually making the LLM call.
|
||||
|
||||
## Session Pinning
|
||||
|
||||
Send an `X-Session-Id` header to pin the routing decision for a session. Once a model is selected, all subsequent requests with the same session ID return the same model without re-running routing.
|
||||
|
||||
```bash
|
||||
# First call — runs routing, caches result
|
||||
curl http://localhost:12000/routing/v1/chat/completions \
|
||||
-H "Content-Type: application/json" \
|
||||
-H "X-Session-Id: my-session-123" \
|
||||
-d '{
|
||||
"model": "gpt-4o-mini",
|
||||
"messages": [{"role": "user", "content": "Write a Python function for binary search"}]
|
||||
}'
|
||||
```
|
||||
|
||||
Response (first call):
|
||||
```json
|
||||
{
|
||||
"model": "anthropic/claude-sonnet-4-20250514",
|
||||
"route": "code_generation",
|
||||
"trace_id": "c16d1096c1af4a17abb48fb182918a88",
|
||||
"session_id": "my-session-123",
|
||||
"pinned": false
|
||||
}
|
||||
```
|
||||
|
||||
```bash
|
||||
# Second call — same session, returns cached result
|
||||
curl http://localhost:12000/routing/v1/chat/completions \
|
||||
-H "Content-Type: application/json" \
|
||||
-H "X-Session-Id: my-session-123" \
|
||||
-d '{
|
||||
"model": "gpt-4o-mini",
|
||||
"messages": [{"role": "user", "content": "Now explain merge sort"}]
|
||||
}'
|
||||
```
|
||||
|
||||
Response (pinned):
|
||||
```json
|
||||
{
|
||||
"model": "anthropic/claude-sonnet-4-20250514",
|
||||
"route": "code_generation",
|
||||
"trace_id": "a1b2c3d4e5f6...",
|
||||
"session_id": "my-session-123",
|
||||
"pinned": true
|
||||
}
|
||||
```
|
||||
|
||||
Session TTL and max cache size are configurable in `config.yaml`:
|
||||
```yaml
|
||||
routing:
|
||||
session_ttl_seconds: 600 # default: 600 (10 minutes)
|
||||
session_max_entries: 10000 # default: 10000
|
||||
```
|
||||
|
||||
Without the `X-Session-Id` header, routing runs fresh every time (no breaking change).
|
||||
|
||||
## Demo Output
|
||||
|
||||
```
|
||||
|
|
@ -88,5 +145,33 @@ The response tells you which model would handle this request and which route was
|
|||
"trace_id": "26be822bbdf14a3ba19fe198e55ea4a9"
|
||||
}
|
||||
|
||||
--- 7. Session pinning - first call (fresh routing decision) ---
|
||||
{
|
||||
"model": "anthropic/claude-sonnet-4-20250514",
|
||||
"route": "code_generation",
|
||||
"trace_id": "f1a2b3c4d5e6f7a8b9c0d1e2f3a4b5c6",
|
||||
"session_id": "demo-session-001",
|
||||
"pinned": false
|
||||
}
|
||||
|
||||
--- 8. Session pinning - second call (same session, pinned) ---
|
||||
Notice: same model returned with "pinned": true, routing was skipped
|
||||
{
|
||||
"model": "anthropic/claude-sonnet-4-20250514",
|
||||
"route": "code_generation",
|
||||
"trace_id": "a9b8c7d6e5f4a3b2c1d0e9f8a7b6c5d4",
|
||||
"session_id": "demo-session-001",
|
||||
"pinned": true
|
||||
}
|
||||
|
||||
--- 9. Different session gets its own fresh routing ---
|
||||
{
|
||||
"model": "openai/gpt-4o",
|
||||
"route": "complex_reasoning",
|
||||
"trace_id": "1a2b3c4d5e6f7a8b9c0d1e2f3a4b5c6d",
|
||||
"session_id": "demo-session-002",
|
||||
"pinned": false
|
||||
}
|
||||
|
||||
=== Demo Complete ===
|
||||
```
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue