Bug
openhuman::memory::sync_pipeline_e2e_tests::multi_batch_volume_builds_full_tree panics with:
should seal to L1+, got max_level=0
The test ingests 30 large chat batches via ingest_chat, drains all jobs, then asserts the source tree sealed to L1+. The seal never fires despite the L0 buffer accumulating enough tokens.
Likely cause: the append_buffer → seal job chain is not triggering correctly after the refactor in #3059 (removed global+topic trees). The drain_until_idle loop exits before the seal cascade propagates.
Repro
cargo test -p openhuman --lib -- sync_pipeline_e2e_tests::multi_batch_volume
Fails on feat/memory-sources-defaults branch and is pre-existing (not introduced by #3113).
Bug
openhuman::memory::sync_pipeline_e2e_tests::multi_batch_volume_builds_full_treepanics with:The test ingests 30 large chat batches via
ingest_chat, drains all jobs, then asserts the source tree sealed to L1+. The seal never fires despite the L0 buffer accumulating enough tokens.Likely cause: the
append_buffer→sealjob chain is not triggering correctly after the refactor in #3059 (removed global+topic trees). Thedrain_until_idleloop exits before the seal cascade propagates.Repro
cargo test -p openhuman --lib -- sync_pipeline_e2e_tests::multi_batch_volumeFails on
feat/memory-sources-defaultsbranch and is pre-existing (not introduced by #3113).