Refactor pool candidate scheduling

This commit is contained in:
fawney19
2026-05-03 20:14:29 +08:00
parent 8ebee9922c
commit a24e4a793d
55 changed files with 4825 additions and 311 deletions
@@ -1,7 +1,9 @@
use tracing::warn;
use crate::ai_serving::planner::candidate_materialization::{
build_local_execution_candidate_attempt_source_with_serving,
materialize_local_execution_candidates_with_serving, LocalCandidateResolutionMode,
LocalExecutionCandidateAttemptSource,
};
use crate::ai_serving::planner::candidate_metadata::{
build_local_execution_candidate_contract_metadata,
@@ -169,3 +171,96 @@ pub(crate) async fn materialize_local_same_format_provider_candidate_attempts(
Ok((outcome.attempts, outcome.candidate_count))
}
pub(crate) async fn build_local_same_format_provider_candidate_attempt_source<'a>(
state: &'a AppState,
trace_id: &str,
input: &LocalSameFormatProviderDecisionInput,
body_json: &serde_json::Value,
spec: LocalSameFormatProviderSpec,
) -> Result<(LocalExecutionCandidateAttemptSource<'a>, usize), GatewayError> {
let spec_metadata = local_same_format_provider_spec_metadata(spec);
let planner_state = PlannerAppState::new(state);
let sticky_session_token = extract_pool_sticky_session_token(body_json);
let persistence_policy = build_local_candidate_persistence_policy(
&input.auth_context,
input.required_capabilities.as_ref(),
LocalCandidatePersistencePolicyKind::SameFormatProviderDecision,
);
let (candidates, preselection_skipped) = planner_state
.list_selectable_candidates_with_skip_reasons(
spec_metadata.api_format,
&input.requested_model,
spec_metadata.require_streaming,
input.required_capabilities.as_ref(),
Some(&input.auth_snapshot),
current_unix_secs(),
)
.await?;
Ok(build_local_execution_candidate_attempt_source_with_serving(
planner_state,
trace_id,
spec_metadata.api_format,
Some(&input.requested_model),
Some(&input.auth_snapshot),
input.required_capabilities.as_ref(),
sticky_session_token.as_deref(),
input.request_auth_channel.as_deref(),
persistence_policy,
candidates,
preselection_skipped
.into_iter()
.map(|item| SkippedLocalExecutionCandidate {
candidate: item.candidate,
skip_reason: item.skip_reason,
transport: None,
ranking: None,
extra_data: None,
})
.collect(),
LocalCandidateResolutionMode::Standard,
|eligible| {
let (execution_strategy, conversion_mode) = ai_local_execution_contract_for_formats(
spec_metadata.api_format,
spec_metadata.api_format,
);
Some(build_local_execution_candidate_contract_metadata(
LocalExecutionCandidateMetadataParts {
eligible,
provider_api_format: spec_metadata.api_format,
client_api_format: spec_metadata.api_format,
extra_fields: serde_json::Map::new(),
},
execution_strategy,
conversion_mode,
spec_metadata.api_format,
))
},
|mut skipped_candidate| {
let provider_api_format = skipped_candidate
.transport
.as_ref()
.map(|transport| transport.endpoint.api_format.trim().to_ascii_lowercase())
.unwrap_or_else(|| spec_metadata.api_format.to_string());
let (execution_strategy, conversion_mode) = ai_local_execution_contract_for_formats(
spec_metadata.api_format,
provider_api_format.as_str(),
);
skipped_candidate.extra_data = Some(
build_local_execution_candidate_contract_metadata_for_candidate(
&skipped_candidate.candidate,
skipped_candidate.transport_ref(),
provider_api_format.as_str(),
spec_metadata.api_format,
serde_json::Map::new(),
execution_strategy,
conversion_mode,
provider_api_format.as_str(),
),
);
skipped_candidate
},
)
.await)
}
@@ -8,10 +8,12 @@ pub(crate) use self::build::{
maybe_build_sync_local_same_format_provider_decision_payload,
};
pub(crate) use self::candidates::{
build_local_same_format_provider_candidate_attempt_source,
materialize_local_same_format_provider_candidate_attempts,
resolve_local_same_format_provider_decision_input,
};
pub(crate) use self::payload::maybe_build_local_same_format_provider_decision_payload_for_candidate;
pub(crate) use crate::ai_serving::planner::candidate_materialization::LocalExecutionCandidateAttempt as LocalSameFormatProviderCandidateAttempt;
pub(crate) use crate::ai_serving::planner::candidate_materialization::LocalExecutionCandidateAttemptSource as LocalSameFormatProviderCandidateAttemptSource;
pub(crate) use crate::ai_serving::planner::decision_input::LocalRequestedModelDecisionInput as LocalSameFormatProviderDecisionInput;
pub(crate) use crate::ai_serving::{LocalSameFormatProviderFamily, LocalSameFormatProviderSpec};