Evict archive writers on deletion and sweep idle archives (closes #89)
All checks were successful
check / check (push) Successful in 3m10s

The per-webhook archiveWriter registry in the database delivery target
was never evicted, so a deleted webhook's writer -- and any archive
file handle open within its debounce window -- lingered for the process
lifetime. Separately, expiry pruning ran only when an archive was
(re)opened, and reopens only happen on writes, so an archive belonging
to a webhook that stopped receiving events kept its expired rows
forever.

Eviction: a new one-method delivery.WebhookEvictor interface (kept
separate from Notifier: archiving lifecycle is not notification) is
implemented by the Engine and injected into the handlers. Deleting a
webhook, or deleting its last database target, drops the writer from
the registry and closes its handle under the writer's own mutex, so
eviction can never race an in-flight write. An evicted writer refuses
further writes rather than reopening a file nothing holds. The archive
file is deliberately left on disk: it is long-term storage an operator
may want to keep or move away, and destroying it as a side effect of
deleting a webhook would be unrecoverable.

Idle sweep: a new ArchiveSweeper, modelled on the event RetentionReaper
(fx lifecycle hooks, cancellable context, WaitGroup, ticker loop),
prunes archives whose database target declares a positive expiry. It
reuses the existing RETENTION_SWEEP_INTERVAL rather than adding a
config key. It never creates an archive -- a missing file is skipped,
and the reopen uses SQLite mode=rw so the file cannot be conjured even
if it disappears mid-sweep -- routes the prune through the per-webhook
writer so its mutex orders the sweep against concurrent writes, and
leaves the archive closed so the move-the-file-away workflow keeps
working. A failure for one webhook is logged and the sweep continues.
Archives with no expiry or the expiry "never" are untouched.
This commit is contained in:
clawbot
2026-08-09 02:25:33 +00:00
parent 4f5ecb18e5
commit 190cabe0f2
15 changed files with 1514 additions and 11 deletions

View File

@@ -24,6 +24,20 @@ const archiveExpiryNever = "never"
// offline archiving, but never more than once per this window.
const archiveReopenDebounce = time.Second
const (
// archiveModeCreate is the SQLite URI mode used by the write
// path: open the archive file, creating it if missing, so a
// first write (or a write after the operator moved the file
// away) recreates it.
archiveModeCreate = "rwc"
// archiveModeExisting is the SQLite URI mode used by the idle
// sweep: open read-write but never create. A sweep must never
// conjure an empty archive file for a webhook that has a
// database target but has never received an event.
archiveModeExisting = "rw"
)
var (
// errArchiveMissingWebhookID is returned when an event to
// archive has no webhook id to key its archive file on.
@@ -44,6 +58,15 @@ var (
errArchiveExpiryNotPositive = errors.New(
"expiry must be a positive duration or \"never\"",
)
// errArchiveWriterEvicted is returned when a writer that has
// been evicted (its webhook was deleted, or its last database
// target was removed) is used again. An evicted writer is no
// longer in the registry, so reopening its file would leak a
// handle nothing owns.
errArchiveWriterEvicted = errors.New(
"archive writer has been evicted",
)
)
// databaseTargetConfig is the optional per-target JSON config
@@ -161,6 +184,12 @@ type archiveWriter struct {
db *gorm.DB
lastReopen time.Time
reopens int
// evicted marks a writer that has been removed from the
// per-webhook registry. Its handle is closed and it must
// never open the file again: nothing holds it any more, so a
// reopen would leak the handle for the process lifetime.
evicted bool
}
// newArchiveWriter builds an archiveWriter for a file path with
@@ -185,6 +214,12 @@ func (w *archiveWriter) write(
w.mu.Lock()
defer w.mu.Unlock()
if w.evicted {
return fmt.Errorf(
"%w: %s", errArchiveWriterEvicted, w.path,
)
}
if w.db == nil || !fileExists(w.path) {
err := w.reopen(expiry)
if err != nil {
@@ -212,7 +247,19 @@ func (w *archiveWriter) write(
// its schema, records the reopen time, and prunes expired rows
// when expiry is positive.
func (w *archiveWriter) open(expiry time.Duration) error {
dbURL := fmt.Sprintf("file:%s?mode=rwc", w.path)
return w.openMode(archiveModeCreate, expiry)
}
// openMode opens the archive file with the given SQLite URI
// mode, migrates its schema, records the reopen time, and
// prunes expired rows when expiry is positive. The write path
// passes archiveModeCreate so a missing file is recreated; the
// idle sweep passes archiveModeExisting so a missing file is an
// error rather than a newly conjured empty archive.
func (w *archiveWriter) openMode(
mode string, expiry time.Duration,
) error {
dbURL := fmt.Sprintf("file:%s?mode=%s", w.path, mode)
sqlDB, err := sql.Open("sqlite", dbURL)
if err != nil {
@@ -275,6 +322,63 @@ func (w *archiveWriter) close() {
w.db = nil
}
// sweepExpired prunes an archive that may have gone idle, with
// no write to trigger the usual on-reopen prune. It takes the
// writer's own mutex for the whole operation, so a sweep is
// ordered against concurrent writes rather than reaching around
// them to the file.
//
// It never creates the archive file: a missing file is skipped,
// and the reopen uses archiveModeExisting so SQLite itself
// refuses to create one if the file disappears between the
// check and the open.
//
// The archive is left CLOSED afterwards. An idle archive holding
// no handle is what keeps the operator's move-the-file-away
// workflow working; the next write reopens (and recreates) the
// file as it always has.
func (w *archiveWriter) sweepExpired(expiry time.Duration) error {
w.mu.Lock()
defer w.mu.Unlock()
if w.evicted {
return fmt.Errorf(
"%w: %s", errArchiveWriterEvicted, w.path,
)
}
if !fileExists(w.path) {
return nil
}
// Drop any live handle first so the prune runs against a
// freshly opened file, matching the write path's semantics.
w.close()
err := w.openMode(archiveModeExisting, expiry)
if err != nil {
return err
}
w.close()
return nil
}
// evict closes the writer's handle and marks it unusable. It is
// called when the writer leaves the registry, either because the
// webhook was deleted or because its last database target was
// removed. The archive FILE is deliberately left on disk: it is
// long-term storage an operator may still want.
func (w *archiveWriter) evict() {
w.mu.Lock()
defer w.mu.Unlock()
w.evicted = true
w.close()
}
// prune deletes archived rows older than expiry, measured from
// each row's archived time. It runs on every (re)open, and
// because the file is reopened after writes this keeps the