Keep EAGLE diagnostics from breaking disagg startup

The C32 startup audit was diagnostic-only, but it incorrectly read spec_algorithm from helper queue objects. PrefillBootstrapQueue and DecodePreallocQueue keep scheduler-owned runtime state instead, so the debug path failed before KV manager initialization could complete.\n\nUse scheduler.spec_algorithm for both startup audits and record the ownership rule in the page-aligned cache ledger.\n\nConstraint: Queue helpers do not copy every scheduler field onto self.\nRejected: Disable EAGLE accept diagnostics | would lose the draft-transfer evidence needed for the accept-len investigation.\nConfidence: high\nScope-risk: narrow\nDirective: Disaggregation helper diagnostics should read scheduler-owned runtime state unless the constructor explicitly copied the field.\nTested: Local py_compile for prefill.py and decode.py.\nTested: Local git diff --check.\nTested: Synced prefill.py/decode.py to g0034 and remote container py_compile passed.\nNot-tested: Full disaggregated startup and ETE traffic after this fix; user controls runtime launch and load generation.

docs/advanced_features/nsa_prefill_cp_page_aligned_cache_contract.md

@@ -1966,3 +1966,20 @@ C32 correction update:
   transfer surfaces are registered.
 - Synced `prefill.py` and `decode.py` to remote `g0034:/mnt/beegfs/cjy/sglang-dev`;
   remote container py_compile passed for the updated files.
+
+### C33. EAGLE accept startup diagnostic must read speculative config from scheduler-owned state
+
+Finding:
+
+- Enabling `SGLANG_EAGLE_ACCEPT_DEBUG=1` caused prefill startup to fail in `PrefillBootstrapQueue._init_kv_manager()` with `AttributeError: 'PrefillBootstrapQueue' object has no attribute 'spec_algorithm'`.
+- The diagnostic added in C32 incorrectly assumed the bootstrap/prealloc queue objects expose `self.spec_algorithm`. They only retain `self.scheduler`; the speculative algorithm lives on `self.scheduler.spec_algorithm` in these queue classes.
+- The same incorrect diagnostic reference existed in `DecodePreallocQueue._init_kv_manager()`, which would have produced the analogous decode startup failure under the same debug env.
+
+Correction:
+
+- Read `self.scheduler.spec_algorithm.is_eagle()` in both prefill and decode KV-manager startup audits.
+- This is a diagnostic-path-only fix. It does not change KV registration, cache ownership, transfer, prefetch, or EAGLE acceptance behavior.
+
+Guardrail:
+
+- Future diagnostics inside disaggregation helper queues should use scheduler-owned runtime state unless the queue constructor explicitly copies that state onto `self`.

python/sglang/srt/disaggregation/decode.py

@@ -509,7 +509,7 @@ class DecodePreallocQueue:
                     len(kv_args.state_data_ptrs),
                 )
 
-        if envs.SGLANG_EAGLE_ACCEPT_DEBUG.get() and self.spec_algorithm.is_eagle():
+        if envs.SGLANG_EAGLE_ACCEPT_DEBUG.get() and self.scheduler.spec_algorithm.is_eagle():
             logger.info(
                 "[EAGLE_ACCEPT_DEBUG] decode_kv_manager cp_rank=%s "
                 "target_kv_bufs=%s draft_kv_bufs=%s total_kv_bufs=%s "

python/sglang/srt/disaggregation/prefill.py

@@ -350,7 +350,7 @@ class PrefillBootstrapQueue:
                     len(kv_args.state_data_ptrs),
                 )
 
-        if envs.SGLANG_EAGLE_ACCEPT_DEBUG.get() and self.spec_algorithm.is_eagle():
+        if envs.SGLANG_EAGLE_ACCEPT_DEBUG.get() and self.scheduler.spec_algorithm.is_eagle():
             logger.info(
                 "[EAGLE_ACCEPT_DEBUG] prefill_kv_manager cp_rank=%s "
                 "target_kv_bufs=%s draft_kv_bufs=%s total_kv_bufs=%s "