Files
Maven/internal/store/delivery.go
T
claude 35c6ff5a71 Make delivery and integration failures explicit
Persist reminder presentations and retry state, atomically complete collapsed deliveries, fall back across away reaches, and block permanent failures visibly (V-715, V-678). Fail closed when enabled integrations lack credentials and keep remote arms explicitly dark (V-691). Give mavweb one sanitized, request-correlated error contract (V-689). Owner explicitly requested direct commits to master.
2026-08-13 02:50:59 +04:00

158 lines
6.4 KiB
Go

package store
import (
"context"
"fmt"
"log"
"time"
)
// DeliveryAttempt statuses. pending = Begin recorded, no Complete yet — either
// still in flight or the process died mid-send (crash window the outbox
// exists to close). sent/failed = Complete recorded the sink's outcome.
// unknown = a pending row found stale at startup: the process that started it
// is gone, and the send may or may not have reached the external channel.
// Never auto-resolved into sent or failed — that would be guessing.
// dropped = the routing table deliberately suppressed this one (a care nudge
// while you're away). Nothing was sent and nothing went wrong; the row exists
// so "she dropped it" and "the rule never fired" don't look the same later.
const (
DeliveryPending = "pending"
DeliverySent = "sent"
DeliveryFailed = "failed"
DeliveryUnknown = "unknown"
DeliveryDropped = "dropped"
)
// BeginDeliveryAttempt durably records intent to send BEFORE the external
// send happens, so a crash between "sent externally" and "recorded" leaves a
// trace instead of silence. kind is "nudge" or "reminder"; rule is set for
// nudges, reminderID for reminders (the other left at its zero value).
// deliveryGroup is the exact persisted reminder occurrence or collapsed
// bundle; it is empty for nudges and legacy reminder attempts.
// bodyHash is an opaque caller-computed key (e.g. sha256 of channel+body) —
// stored for post-crash operator triage, not enforced as a uniqueness
// constraint (a rule/reminder legitimately re-sends across ticks).
func (s *Store) BeginDeliveryAttempt(ctx context.Context, kind, rule string, reminderID int64, deliveryGroup, channel, bodyHash string, now time.Time) (int64, error) {
res, err := s.db.ExecContext(ctx,
`INSERT INTO delivery_attempts (kind, rule, reminder_id, delivery_group, channel, body_hash, status, created_ts)
VALUES (?, ?, ?, ?, ?, ?, 'pending', ?)`,
kind, rule, reminderID, deliveryGroup, channel, bodyHash, now.UnixMilli())
if err != nil {
return 0, fmt.Errorf("begin delivery attempt: %w", err)
}
id, err := res.LastInsertId()
if err != nil {
return 0, fmt.Errorf("begin delivery attempt: last insert id: %w", err)
}
return id, nil
}
// CompleteDeliveryAttempt records the sink's outcome for a prior
// BeginDeliveryAttempt. status is "sent", "failed" or "dropped" — never
// "pending" or "unknown" (those are set only by Begin and reconciliation
// respectively).
func (s *Store) CompleteDeliveryAttempt(ctx context.Context, id int64, status string, now time.Time) error {
if status != DeliverySent && status != DeliveryFailed && status != DeliveryDropped {
return fmt.Errorf("store: invalid delivery completion status %q", status)
}
_, err := s.db.ExecContext(ctx,
`UPDATE delivery_attempts SET status = ?, completed_ts = ? WHERE id = ? AND status = 'pending'`,
status, now.UnixMilli(), id)
if err != nil {
return fmt.Errorf("complete delivery attempt %d: %w", id, err)
}
return nil
}
// ReconcileStaleDeliveryAttempts runs once at daemon startup, before the tick
// loop resumes sending. Any attempt still "pending" from a previous process
// life is the exact crash window the outbox exists to close: the external
// send may have landed and the process died before recording the outcome.
// Marking it "unknown" (rather than silently resending, and rather than
// silently dropping it) preserves the same never-guess-an-ambiguous-outcome
// rule as the IPC client and Hexis execution engine. Returns the count
// reconciled, for startup logging.
func (s *Store) ReconcileStaleDeliveryAttempts(ctx context.Context, now time.Time) (int, error) {
res, err := s.db.ExecContext(ctx,
`UPDATE delivery_attempts SET status = 'unknown', completed_ts = ? WHERE status = 'pending'`,
now.UnixMilli())
if err != nil {
return 0, fmt.Errorf("reconcile stale delivery attempts: %w", err)
}
n, err := res.RowsAffected()
if err != nil {
return 0, fmt.Errorf("reconcile stale delivery attempts: rows affected: %w", err)
}
if n > 0 {
log.Printf("store: reconciled %d stale delivery attempt(s) from a prior run as outcome=unknown", n)
}
return int(n), nil
}
// DeliveryAttempt — one row of the outbox, as a reader sees it.
type DeliveryAttempt struct {
ID int64
Kind string // nudge|reminder
Rule string // set for nudges
ReminderID int64 // set for reminders
DeliveryGroup string // exact reminder occurrence/bundle; empty for nudges and legacy rows
Channel string
Status string // one of the Delivery* constants
Created time.Time
Completed time.Time // zero while pending
HasComplete bool
}
// ListDeliveryAttempts returns recent attempts, newest first. An empty status
// means every status; anything else filters on it.
//
// The table was write-only until 04-08-2026: rows were recorded and nothing
// could read them, so the tests for #368 and #370 had to reach past the store
// into store.DB, which is the tell (Vikunja #390). A durable record nobody can
// read answers no question, and "why did Maven go quiet" is supposed to be a
// query rather than a mystery.
//
// Status is the filter that earns its place, because the two questions actually
// asked are "what got dropped" and "what is still pending". Neither is
// answerable by reading the whole list on a busy day.
func (s *Store) ListDeliveryAttempts(ctx context.Context, status string, limit int) ([]DeliveryAttempt, error) {
if limit <= 0 {
limit = 50
}
q := `SELECT id, kind, rule, reminder_id, delivery_group, channel, status, created_ts, completed_ts
FROM delivery_attempts`
args := []any{}
if status != "" {
q += ` WHERE status = ?`
args = append(args, status)
}
q += ` ORDER BY created_ts DESC, id DESC LIMIT ?`
args = append(args, limit)
rows, err := s.db.QueryContext(ctx, q, args...)
if err != nil {
return nil, fmt.Errorf("list delivery attempts: %w", err)
}
defer rows.Close()
var out []DeliveryAttempt
for rows.Next() {
var a DeliveryAttempt
var created int64
var completed *int64
if err := rows.Scan(&a.ID, &a.Kind, &a.Rule, &a.ReminderID, &a.DeliveryGroup, &a.Channel, &a.Status, &created, &completed); err != nil {
return nil, fmt.Errorf("list delivery attempts: scan: %w", err)
}
a.Created = time.UnixMilli(created)
if completed != nil {
a.Completed, a.HasComplete = time.UnixMilli(*completed), true
}
out = append(out, a)
}
if err := rows.Err(); err != nil {
return nil, fmt.Errorf("list delivery attempts: %w", err)
}
return out, nil
}