i/Aether
1
0
Fork 0
mirror of https://github.com/fawney19/Aether.git synced 2026-09-03 01:40:21 +08:00
Code Issues Packages Projects Releases Wiki Activity

feat: pass session affinity through planner inputs

Browse Source
This commit is contained in:
RWDai
2026-05-05 11:23:54 +08:00
parent 37da4e245a
commit bcd542f7ee
3 changed files with 19 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
apps/aether-gateway/src/ai_serving/planner/state/scheduler.rs
View File

@@ -1,4 +1,4 @@
use aether_scheduler_core::SchedulerMinimalCandidateSelectionCandidate;
use aether_scheduler_core::{ClientSessionAffinity, SchedulerMinimalCandidateSelectionCandidate};
use std::time::Duration;
use tokio::time::Instant;
@@ -18,6 +18,7 @@ impl<'a> PlannerAppState<'a> {
require_streaming: bool,
required_capabilities: Option<&serde_json::Value>,
auth_snapshot: Option<&GatewayAuthApiKeySnapshot>,
client_session_affinity: Option<&ClientSessionAffinity>,
now_unix_secs: u64,
) -> Result<Vec<SchedulerMinimalCandidateSelectionCandidate>, GatewayError> {
let enable_model_directives =
@@ -35,6 +36,7 @@ impl<'a> PlannerAppState<'a> {
require_streaming,
required_capabilities,
auth_snapshot,
client_session_affinity,
now_unix_secs,
enable_model_directives,
)
@@ -48,6 +50,7 @@ impl<'a> PlannerAppState<'a> {
require_streaming: bool,
required_capabilities: Option<&serde_json::Value>,
auth_snapshot: Option<&GatewayAuthApiKeySnapshot>,
client_session_affinity: Option<&ClientSessionAffinity>,
now_unix_secs: u64,
) -> Result<
(
@@ -77,6 +80,7 @@ impl<'a> PlannerAppState<'a> {
require_streaming,
required_capabilities,
auth_snapshot,
client_session_affinity,
attempt_now_unix_secs,
enable_model_directives,
)
@@ -106,6 +110,7 @@ impl<'a> PlannerAppState<'a> {
candidates: Vec<SchedulerMinimalCandidateSelectionCandidate>,
required_capabilities: Option<&serde_json::Value>,
auth_snapshot: Option<&GatewayAuthApiKeySnapshot>,
client_session_affinity: Option<&ClientSessionAffinity>,
now_unix_secs: u64,
) -> Result<
(
@@ -121,6 +126,7 @@ impl<'a> PlannerAppState<'a> {
candidates,
required_capabilities,
auth_snapshot,
client_session_affinity,
now_unix_secs,
)
.await