[L-01 dup?] hero_db + hero_indexer show 'failed' in proc service list after clean --download start #65

New issue

Open

opened 2026-05-07 17:29:30 +00:00 by mik-tf · 0 comments

mik-tf commented

2026-05-07 17:29:30 +00:00

Owner

Surfaced during session 74 from-nothing-on-DO validation.

Symptom

service_db install --download --version v0.3.2 --reset and service_indexer install --download --version v0.1.3 --reset both completed successfully (✓ registered & started). hero_proc dashboard immediately after showed both as failed while the other 12 services were running.

Possible explanations

L-01 status drift (hero_proc#83) — hero_proc reports stale state after restart. The actual processes may be running fine.
Real runtime crash — services start, then exit on first request or during init.
Missing dep — db needs Redis (hero_collab triggered a 'Redis not listening on 127.0.0.1:6379' warning during the run). hero_db itself bundles Redis, but maybe the bundled instance failed to start.

Reproduction (session 75)

nu -c "use nutools/modules/services/service_db.nu; service_db start --download --reset"
sleep 5
hero_proc service status hero_db
hero_proc logs tail hero_db_server --lines 100
ps -ef | grep hero_db

Same for hero_indexer. If logs show the process is running but hero_proc reports failed → L-01. If logs show a crash → new bug, file separately.

Severity

Medium-high — both services are core (db underpins hero_books / persistence; indexer underpins agent grounding). If the drift is real, demo functionality breaks even though the bootstrap pipeline works.

Filed as part of home#227 follow-ups. Linked to L-01 hero_proc status drift.

Surfaced during session 74 from-nothing-on-DO validation. ## Symptom `service_db install --download --version v0.3.2 --reset` and `service_indexer install --download --version v0.1.3 --reset` both completed successfully (✓ registered & started). hero_proc dashboard immediately after showed both as `failed` while the other 12 services were `running`. ## Possible explanations 1. **L-01 status drift** ([hero_proc#83](https://forge.ourworld.tf/lhumina_code/hero_proc/issues/83)) — hero_proc reports stale state after restart. The actual processes may be running fine. 2. **Real runtime crash** — services start, then exit on first request or during init. 3. **Missing dep** — db needs Redis (hero_collab triggered a 'Redis not listening on 127.0.0.1:6379' warning during the run). hero_db itself bundles Redis, but maybe the bundled instance failed to start. ## Reproduction (session 75) ```bash nu -c "use nutools/modules/services/service_db.nu; service_db start --download --reset" sleep 5 hero_proc service status hero_db hero_proc logs tail hero_db_server --lines 100 ps -ef | grep hero_db ``` Same for hero_indexer. If logs show the process is running but hero_proc reports failed → L-01. If logs show a crash → new bug, file separately. ## Severity Medium-high — both services are core (db underpins hero_books / persistence; indexer underpins agent grounding). If the drift is real, demo functionality breaks even though the bootstrap pipeline works. Filed as part of [home#227](https://forge.ourworld.tf/lhumina_code/home/issues/227) follow-ups. Linked to [L-01 hero_proc status drift](../limitations/L-01-hero_proc-status-drift.md).