Commit Graph

108 Commits

Author SHA1 Message Date
b4ef1de30f feat(graph_engine): add ready_queue state persistence to GraphRuntimeState
- Add ReadyQueueState TypedDict for type-safe queue serialization
- Add ready_queue attribute to GraphRuntimeState for initializing with pre-existing queue state
- Update GraphEngine to load ready_queue from GraphRuntimeState on initialization
- Implement proper type hints using ReadyQueueState for better type safety
- Add comprehensive tests for ready_queue loading functionality

The ready_queue is read-only after initialization and allows resuming workflow
execution with a pre-populated queue of nodes ready to execute.
2025-09-15 03:05:10 +08:00
4cdc19fd05 feat(graph_engine): add abstract layer and dump / load methods for ready queue. 2025-09-14 04:19:24 +08:00
7e69403dda refactor(graph_engine): use singledispatchmethod in event_handler
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-10 03:12:33 +08:00
836ed1f380 refactor(graph_engine): Move ErrorHandler into a single file package
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-10 02:35:05 +08:00
f17c71e08a refactor(graph_engine): Move GraphStateManager to single file package.
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-10 01:55:30 +08:00
d52621fce3 refactor(graph_engine): Merge error strategies into error_handler.py
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-10 01:49:46 +08:00
a23c8fcb1a refactor: move execution limits from engine core to layer
Remove max_execution_time and max_execution_steps from ExecutionContext and GraphEngine since these limits are now handled by ExecutionLimitsLayer. This follows the separation of concerns principle by keeping execution limits as a cross-cutting concern handled by layers rather than embedded in core engine components.

Changes:
- Remove max_execution_time and max_execution_steps from ExecutionContext
- Remove these parameters from GraphEngine.__init__()
- Remove max_execution_time from Dispatcher
- Update workflow_entry.py to no longer pass these parameters
- Update all tests to remove these parameters
2025-09-10 01:32:45 +08:00
103a9a4e67 fix(graph_engine): add type hint for workers_to_remove 2025-09-05 01:59:11 +08:00
15b3443e9e fix(debug_logging_layer): remove access for variable pool 2025-09-05 01:52:19 +08:00
e229510e73 perf: eliminate lock contention in worker pool by removing callbacks
Remove worker idle/active callbacks that caused severe lock contention.
Instead, use sampling-based monitoring where worker states are queried
on-demand during scaling decisions. This eliminates the performance
bottleneck caused by workers acquiring locks 10+ times per second.

Changes:
- Remove callback parameters from Worker class
- Add properties to expose worker idle state directly
- Update WorkerPool to query worker states without callbacks
- Maintain scaling functionality with better performance
2025-09-04 19:37:31 +08:00
36048d1526 feat(graph_engine): allow to scale down without lock
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-04 19:32:07 +08:00
ad9eed2551 fix: disable scale for perfermance
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-04 19:11:22 +08:00
4d63bd2083 refactor(graph_engine): rename SimpleWorkerPool to WorkerPool 2025-09-04 17:47:13 +08:00
226f14a20f feat(graph_engine): implement scale down worker
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-04 15:35:20 +08:00
04bbf540d9 chore: code format
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-04 02:33:53 +08:00
657c27ec75 feat(graph_engine): make runtime state read-only in layer
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-04 02:30:40 +08:00
8332472944 refactor(graph_engine): rename Layer to GraphEngineLayer
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-04 02:11:31 +08:00
fe3f03e50a feat: add property-based access control to GraphRuntimeState
- Replace direct field access with private attributes and property decorators
- Implement deep copy protection for mutable objects (dict, LLMUsage)
- Add helper methods: set_output(), get_output(), update_outputs()
- Add increment_node_run_steps() and add_tokens() convenience methods
- Update loop_node and event_handlers to use new accessor methods
- Add comprehensive unit tests for immutability and validation
- Ensure backward compatibility with existing property access patterns
2025-09-04 02:08:58 +08:00
1770b93e5b chore(graph_engine): Add a TODO commment in _update_response_outputs in event_handlers
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-02 15:20:03 +08:00
d8ff4aa9ba feat(graph_engine): Handle NodeRunAgentLogEvent
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-02 15:02:07 +08:00
9f8f21bf87 chore: remove backup files
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-02 15:01:58 +08:00
0b0dc63f29 Merge remote-tracking branch 'origin/main' into feat/queue-based-graph-engine 2025-09-02 11:52:25 +08:00
ffba341258 [CHORE]: remove redundant-cast (#24807) 2025-09-01 14:05:32 +08:00
8433cf4437 refactor(graph_engine): Merge event_collector and event_emitter into event_manager
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-01 13:15:58 +08:00
bb5d52539c refactor(graph_engine): Merge branch_handler into edge_processor
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-01 12:53:06 +08:00
88622f70fb refactor(graph_engine): Move setup methods into __init__
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-01 12:08:03 +08:00
0fdb1b2bc9 refactor(graph_engine): Correct private attributes and private methods naming
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-01 04:37:23 +08:00
a5cb9d2b73 refactor(graph_engine): inline output_registry into response_coordinator
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-01 03:59:53 +08:00
64c1234724 refactor(graph_engine): Merge worker management into one WorkerPool
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-01 03:23:47 +08:00
202fdfcb81 refactor(graph_engine): Remove backward compatibility code
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-01 02:41:16 +08:00
e2f4c9ba8d refactor(graph_engine): Merge state managers into unified_state_manager
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-09-01 02:08:08 +08:00
82193580de chore: improve typing
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-30 16:35:57 +08:00
c396788128 chore(graph_engine): add final mark to classes
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 15:38:35 +08:00
5c01dd97e8 clean typos words. (#24667)
Signed-off-by: zhanluxianshen <zhanluxianshen@163.com>
2025-08-28 15:23:59 +08:00
e3a7b1f691 fix: type hints
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 05:24:18 +08:00
8aab7f49c3 chore(graph_engine): Use XXX | None instead of Optional[XXX] 2025-08-28 05:09:33 +08:00
1e12c1cbf2 [autofix.ci] apply automated fixes 2025-08-27 21:00:36 +00:00
affedd6ce4 chore(graph_engine): Use XXX | None instead of Optional[XXX] 2025-08-28 04:59:49 +08:00
ef21097774 refactor(graph_engine): Remove unnecessary check from SkipPropagator
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 04:45:26 +08:00
1d377fe994 refactor(graph_engine): Use _ to mark unused variable in BranchHandler
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 04:44:45 +08:00
c82697f267 refactor(graph_engine): Remove node_id from SkipPropagator.skip_branch_paths
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 04:43:56 +08:00
98b25c0bbc refactor(graph_engine): Convert attrs to private in error_handler
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 04:42:37 +08:00
1cd0792606 chore(graph_events): Improve type hints
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 04:41:48 +08:00
7cbf4093f4 chore(graph_engine): Use TYPE | None instead of Optional
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 04:30:50 +08:00
8129ca7c05 chore(graph_engine): Move error_strategy.py to protocols/
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 04:29:32 +08:00
65617f000d feat(event_collector): Update to use ReadWriteLock 2025-08-28 03:26:42 +08:00
55085a9ca2 chore(graph_engine): add type hint for event_queue
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-28 02:38:56 +08:00
48cbf4c78f [autofix.ci] apply automated fixes 2025-08-27 15:33:30 +08:00
8c35663220 feat: queue-based graph engine
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-27 15:33:28 +08:00
da9af7b547 [Chore/Refactor] Use centralized naive_utc_now for UTC datetime operations (#24352)
Signed-off-by: -LAN- <laipz8200@outlook.com>
2025-08-22 23:53:05 +08:00