All checks were successful
check / check (push) Successful in 3m3s
An operator running `sqlite3 <db> .dump` against their own per-webhook database wedged it: inbound webhooks rejected with HTTP 500, delivered webhooks stranded at `pending`, and every one of them POSTed a second time on the next restart while the event log recorded a single attempt. Durability. Every SQLite file — main, per-webhook, and archive — now opens through one path, `internal/database/sqlite_open.go`, in WAL journal mode with a 10-second busy timeout, `BEGIN IMMEDIATE` transactions, and a bounded connection pool. WAL is what stops a reader blocking writers at all. `_txlock=immediate` is what stops a `COMMIT` failing while its transaction stays open on a pooled connection, which is how four `database is locked` errors became 593 `cannot start a transaction within a transaction`. `cache=shared` is gone, because under it an in-process conflict is SQLITE_LOCKED, which the busy handler does not retry. The busy timeout is applied before journal_mode: the driver runs DSN pragmas in order on every new connection, and `PRAGMA journal_mode` takes a lock, so the reverse order leaves the one pragma that can block uncovered by the handler meant to cover it. Eligibility. `internal/delivery/inflight.go` holds the set of deliveries the engine owns — taken when a task is queued, when a target schedules a retry, and by every recovery path before it re-dispatches; dropped when the worker that ran the task returns. Recovery and both sweep arms re-dispatch only what the set does not hold. Nothing decides that from a row's age: a delivery waiting in a 10000-deep channel is arbitrarily old and perfectly healthy, and reasoning from age re-sends it. `takeForRedispatch` is the single gate every re-dispatch goes through — ownership first, then a conditional update confirming the row is still in the status the batch read. Bookkeeping. `recordResult` and `updateDeliveryStatus` return their errors instead of logging and dropping them, and a caller whose bookkeeping write failed writes nothing at all: the delivery keeps whichever non-terminal status it already held, and the sweeps recover it. Every recovery path — pending and retrying alike — first settles any delivery that already holds a successful `DeliveryResult` rather than sending it again. Recovery continues each delivery's own attempt numbering instead of restarting at 1. The sweep gains a `pending`-with-age-bound arm, so a stranded delivery no longer waits for a restart. Docs. WAL produces `-wal`/`-shm` sidecars, so the backup and restore procedures in README.md are corrected against measurement: both documented procedures were re-run against a live instance, a `-wal` left by a crash carries data the `.db` alone does not, and an archive file normally holds its rows in a `-wal` rather than in the `.db`.
316 lines
8.1 KiB
Go
316 lines
8.1 KiB
Go
package delivery
|
|
|
|
import (
|
|
"context"
|
|
"fmt"
|
|
"path/filepath"
|
|
"sync"
|
|
"time"
|
|
|
|
"gorm.io/gorm"
|
|
"sneak.berlin/go/webhooker/internal/database"
|
|
)
|
|
|
|
// databaseTarget is a no-retry target that archives the
|
|
// full inbound event into a per-webhook archive SQLite file,
|
|
// separate from the per-webhook event database. The event is
|
|
// already persisted in the per-webhook event DB by the time
|
|
// delivery runs; the database target additionally writes a
|
|
// durable long-term copy into archive-{webhookID}.db and then
|
|
// records a single attempt whose outcome reflects whether the
|
|
// archive write succeeded. See archiveWriter for the
|
|
// close/reopen, auto-recreate, and expiry semantics.
|
|
type databaseTarget struct {
|
|
eng *Engine
|
|
|
|
mu sync.Mutex
|
|
writers map[string]*archiveWriter
|
|
}
|
|
|
|
// Deliver implements Target. It archives the event, then
|
|
// records one successful attempt and marks the delivery
|
|
// delivered. An archiving error fails the delivery: the
|
|
// attempt is recorded as failed with the error and the
|
|
// delivery is marked failed, so a target that could not do
|
|
// its one job (archiving) never reports success. The target
|
|
// does not retry; the event remains durably stored in the
|
|
// per-webhook event database.
|
|
func (t *databaseTarget) Deliver(
|
|
_ context.Context,
|
|
webhookDB *gorm.DB,
|
|
d *database.Delivery,
|
|
_ *Task,
|
|
_ Scheduler,
|
|
) {
|
|
start := time.Now()
|
|
|
|
err := t.archive(d)
|
|
|
|
elapsed := time.Since(start)
|
|
|
|
t.eng.observeAttempt(d.Target.Type, elapsed)
|
|
|
|
if err != nil {
|
|
t.eng.log.Error(
|
|
"failed to archive event to database target",
|
|
"delivery_id", d.ID,
|
|
"event_id", d.EventID,
|
|
"error", err,
|
|
)
|
|
|
|
recErr := t.eng.recordResult(
|
|
webhookDB, d, 1, false, 0, "",
|
|
err.Error(), elapsed.Milliseconds(),
|
|
)
|
|
if recErr != nil {
|
|
t.eng.bookkeepingFailed(d, recErr)
|
|
|
|
return
|
|
}
|
|
|
|
t.eng.settleStatus(
|
|
webhookDB, d, d.Target.Type,
|
|
database.DeliveryStatusFailed,
|
|
)
|
|
|
|
return
|
|
}
|
|
|
|
recErr := t.eng.recordResult(
|
|
webhookDB, d, 1, true, 0, "", "",
|
|
elapsed.Milliseconds(),
|
|
)
|
|
if recErr != nil {
|
|
t.eng.bookkeepingFailed(d, recErr)
|
|
|
|
return
|
|
}
|
|
|
|
t.eng.settleStatus(
|
|
webhookDB, d, d.Target.Type,
|
|
database.DeliveryStatusDelivered,
|
|
)
|
|
}
|
|
|
|
// archive writes the full event as a row into the webhook's
|
|
// archive database, honouring the optional per-target expiry
|
|
// parsed from the target config JSON.
|
|
func (t *databaseTarget) archive(d *database.Delivery) error {
|
|
webhookID := d.Event.WebhookID
|
|
if webhookID == "" {
|
|
return errArchiveMissingWebhookID
|
|
}
|
|
|
|
expiry, err := parseArchiveExpiry(d.Target.Config)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
w, err := t.writerFor(webhookID)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
row := archivedEvent{
|
|
EventID: d.Event.ID,
|
|
WebhookID: webhookID,
|
|
EntrypointID: d.Event.EntrypointID,
|
|
Method: d.Event.Method,
|
|
Headers: d.Event.Headers,
|
|
Body: d.Event.Body,
|
|
ContentType: d.Event.ContentType,
|
|
}
|
|
|
|
return w.write(row, expiry)
|
|
}
|
|
|
|
// writerFor returns the archiveWriter for a webhook, creating
|
|
// and caching it on first use. Each webhook has one writer so
|
|
// its close/reopen debounce state is shared across concurrent
|
|
// deliveries. The archive file lives beside the per-webhook
|
|
// event database in the data directory.
|
|
func (t *databaseTarget) writerFor(
|
|
webhookID string,
|
|
) (*archiveWriter, error) {
|
|
path, err := t.archivePath(webhookID)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
t.mu.Lock()
|
|
defer t.mu.Unlock()
|
|
|
|
if t.writers == nil {
|
|
t.writers = make(map[string]*archiveWriter)
|
|
}
|
|
|
|
w, ok := t.writers[webhookID]
|
|
if !ok {
|
|
w = newArchiveWriter(path, t.eng.log)
|
|
t.writers[webhookID] = w
|
|
}
|
|
|
|
// A delivery claims the entry: even if the idle sweep created
|
|
// it moments ago, it now belongs to the registry proper and
|
|
// the sweep must leave it in place when it finishes.
|
|
w.sweepOwned = false
|
|
|
|
return w, nil
|
|
}
|
|
|
|
// sweepWriterFor returns the archive writer the idle sweep should
|
|
// prune a webhook through, together with whether the sweep itself
|
|
// created the registry entry.
|
|
//
|
|
// The sweep must route its prune through the registered writer so
|
|
// the writer's mutex orders it against concurrent writes, but it
|
|
// must never leave a registry entry behind: a sweep that ran
|
|
// concurrently with the webhook's deletion would otherwise
|
|
// re-create an entry that nothing will ever evict again, which is
|
|
// exactly the leak eviction exists to prevent. An entry the sweep
|
|
// creates is therefore marked sweep-owned and handed back to
|
|
// releaseSweepWriter when the sweep is done.
|
|
func (t *databaseTarget) sweepWriterFor(
|
|
webhookID string,
|
|
) (*archiveWriter, bool, error) {
|
|
path, err := t.archivePath(webhookID)
|
|
if err != nil {
|
|
return nil, false, err
|
|
}
|
|
|
|
t.mu.Lock()
|
|
defer t.mu.Unlock()
|
|
|
|
if t.writers == nil {
|
|
t.writers = make(map[string]*archiveWriter)
|
|
}
|
|
|
|
w, ok := t.writers[webhookID]
|
|
if ok {
|
|
return w, false, nil
|
|
}
|
|
|
|
w = newArchiveWriter(path, t.eng.log)
|
|
w.sweepOwned = true
|
|
t.writers[webhookID] = w
|
|
|
|
return w, true, nil
|
|
}
|
|
|
|
// releaseSweepWriter drops a registry entry that the idle sweep
|
|
// created, so a sweep leaves the registry exactly as it found it.
|
|
//
|
|
// The entry is removed only if it is still the very writer the
|
|
// sweep installed and no delivery has claimed it in the meantime
|
|
// (writerFor clears sweepOwned when it hands a writer to the
|
|
// write path). Both conditions are evaluated under the registry
|
|
// lock, so an eviction that raced the sweep — which removes the
|
|
// entry outright — simply finds nothing left to do here, and a
|
|
// delivery that adopted the writer keeps a registered, evictable
|
|
// one.
|
|
func (t *databaseTarget) releaseSweepWriter(
|
|
webhookID string, w *archiveWriter,
|
|
) {
|
|
t.mu.Lock()
|
|
defer t.mu.Unlock()
|
|
|
|
cur, ok := t.writers[webhookID]
|
|
if !ok || cur != w || !cur.sweepOwned {
|
|
return
|
|
}
|
|
|
|
delete(t.writers, webhookID)
|
|
}
|
|
|
|
// archivePath returns the archive file path for a webhook: it
|
|
// lives beside the per-webhook event database in the data
|
|
// directory. It does not touch the filesystem.
|
|
func (t *databaseTarget) archivePath(
|
|
webhookID string,
|
|
) (string, error) {
|
|
if t.eng.dbManager == nil {
|
|
return "", errArchiveNoDataDir
|
|
}
|
|
|
|
dir := filepath.Dir(t.eng.dbManager.DBPath(webhookID))
|
|
|
|
return filepath.Join(
|
|
dir, fmt.Sprintf("archive-%s.db", webhookID),
|
|
), nil
|
|
}
|
|
|
|
// evict drops a webhook's archive writer from the registry and
|
|
// closes its handle, so a deleted webhook does not leave a
|
|
// writer (and an open archive handle within its debounce
|
|
// window) alive for the process lifetime.
|
|
//
|
|
// The map entry is removed under the registry lock, which is
|
|
// then released before the handle is closed under the writer's
|
|
// own lock: that ordering keeps the registry available to other
|
|
// webhooks while an in-flight write on this one drains, and
|
|
// closing under the writer's lock means eviction can never race
|
|
// a write.
|
|
//
|
|
// Eviction is idempotent and silent for a webhook with no
|
|
// writer, which is the common case: a webhook with no database
|
|
// target never creates one. It never deletes the archive file.
|
|
func (t *databaseTarget) evict(webhookID string) {
|
|
t.mu.Lock()
|
|
|
|
w, ok := t.writers[webhookID]
|
|
if ok {
|
|
delete(t.writers, webhookID)
|
|
}
|
|
|
|
t.mu.Unlock()
|
|
|
|
if !ok {
|
|
return
|
|
}
|
|
|
|
w.evict()
|
|
|
|
t.eng.log.Info(
|
|
"evicted archive writer",
|
|
"webhook_id", webhookID,
|
|
"path", w.path,
|
|
)
|
|
}
|
|
|
|
// sweepWebhook prunes one webhook's archive of rows older than
|
|
// expiry, without requiring a write. It returns nil (nothing to
|
|
// do) when the archive file does not exist, so a sweep never
|
|
// creates an archive for a webhook that has a database target
|
|
// but has never received an event.
|
|
//
|
|
// It also never leaves a registry entry behind: an entry it had
|
|
// to create to reach the writer's mutex is released again once
|
|
// the prune is done, so a sweep racing a webhook deletion cannot
|
|
// resurrect the writer the eviction just dropped.
|
|
func (t *databaseTarget) sweepWebhook(
|
|
webhookID string, expiry time.Duration,
|
|
) error {
|
|
path, err := t.archivePath(webhookID)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
// Check before taking a writer at all: a webhook whose
|
|
// archive has never been created gets no writer, no handle,
|
|
// and no file.
|
|
if !fileExists(path) {
|
|
return nil
|
|
}
|
|
|
|
w, created, err := t.sweepWriterFor(webhookID)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
if created {
|
|
defer t.releaseSweepWriter(webhookID, w)
|
|
}
|
|
|
|
return w.sweepExpired(expiry)
|
|
}
|