Maddox
eb9fc07021
The v0.4 schema migration silently failed on first deploy — Maubot's UpgradeTable hook didn't create the subscriptions / digest_state tables in the shared Postgres instance. When the digest fired at 8 AM, the SELECT against digest_state raised, the loop's only except was CancelledError, and the task crashed silently. No digest, no logs. Two fixes: - _ensure_db_schema() runs CREATE TABLE IF NOT EXISTS on start() so the bot self-heals if migrations don't run for any reason - digest loop's inner work is now wrapped — any exception sleeps an hour and retries instead of killing the whole schedule |
||
|---|---|---|
| .. | ||
| clients | ||
| __init__.py | ||
| bot.py | ||
| db.py | ||