Retry failed GitHub release scans with backoff (#133)

Releasing a GitHub sync lease always advanced `last_synced_at`, even after a failed scan (e.g. a rate-limit 403). A remote that failed once waited a full `mutable_ttl` before retrying, and a cold remote kept returning 503 until then.

- record scan outcomes in one shared lease helper for github_rpm/deb/alpine
- keep `last_synced_at` and the ETag on failure; retry from 60s with exponential backoff, capped at min(10m, ttl/4)
- honour `Retry-After` / `X-RateLimit-Reset`, clamped to `mutable_ttl`
- add `sync_failures` / `next_retry_at` columns (migration 0002)

Reviewed-on: #133
Co-authored-by: unkin-agent <unkin-agent@unkin.net>
Co-committed-by: unkin-agent <unkin-agent@unkin.net>

internal/database/github_sync.go

@@ -7,6 +7,8 @@ import (
 
 	"github.com/jackc/pgx/v5"
 
+	"git.unkin.net/unkin/artifactapi/internal/provider"
+
 	"git.unkin.net/unkin/artifactapi/pkg/models"
 )
 
@@ -30,22 +32,35 @@ func (db *DB) ListGitHubRPMRemotes(ctx context.Context) ([]models.Remote, error)
 	return remotes, rows.Err()
 }
 
-// ClaimGitHubSyncLease atomically claims the per-remote sync lease. It succeeds
-// (claimed=true) only when the remote is due — never synced, or synced longer
-// than freshness ago — and no live lease is held by another replica. This bounds
-// total GitHub load to roughly one scan per freshness window regardless of how
-// many replicas poll. The returned etag is the stored releases-list ETag, shared
-// across replicas so a conditional request can short-circuit an unchanged repo.
-// A zero freshness (used for prime scans) ignores the recency gate and claims
-// whenever no live lease is held.
+// ClaimGitHubSyncLease atomically claims the per-remote github_rpm sync lease.
+// See claimSyncLease.
 func (db *DB) ClaimGitHubSyncLease(ctx context.Context, remoteName, owner string, freshness, lease time.Duration) (bool, string, error) {
+	return db.claimSyncLease(ctx, "github_rpm_sync_state", remoteName, owner, freshness, lease)
+}
+
+// ReleaseGitHubSyncLease records a github_rpm scan outcome and frees the lease.
+// See releaseSyncLease.
+func (db *DB) ReleaseGitHubSyncLease(ctx context.Context, remoteName, owner string, res provider.SyncResult) error {
+	return db.releaseSyncLease(ctx, "github_rpm_sync_state", remoteName, owner, res)
+}
+
+// claimSyncLease atomically claims a per-remote sync lease in table. It succeeds
+// (claimed=true) only when the remote is due and no live lease is held by
+// another replica. Due means: a pending retry after a failed scan has reached
+// next_retry_at, or, with no retry pending, the remote was never synced or was
+// synced longer than freshness ago. A zero freshness (prime scans) ignores the
+// recency gate but still honours a pending retry's backoff. The returned etag is
+// the stored releases-list ETag, shared across replicas so a conditional request
+// can short-circuit an unchanged repo.
+func (db *DB) claimSyncLease(ctx context.Context, table, remoteName, owner string, freshness, lease time.Duration) (bool, string, error) {
 	row := db.Pool.QueryRow(ctx, `
-		INSERT INTO github_rpm_sync_state AS s (remote_name, sync_lease_owner, sync_lease_expires)
+		INSERT INTO `+table+` AS s (remote_name, sync_lease_owner, sync_lease_expires)
 		VALUES ($1, $2, now() + make_interval(secs => $4))
 		ON CONFLICT (remote_name) DO UPDATE
 			SET sync_lease_owner   = $2,
 			    sync_lease_expires = now() + make_interval(secs => $4)
-			WHERE (s.last_synced_at IS NULL OR s.last_synced_at < now() - make_interval(secs => $3))
+			WHERE (CASE WHEN s.next_retry_at IS NOT NULL THEN s.next_retry_at <= now()
+			            ELSE s.last_synced_at IS NULL OR s.last_synced_at < now() - make_interval(secs => $3) END)
 			  AND (s.sync_lease_expires IS NULL OR s.sync_lease_expires < now())
 		RETURNING s.etag
 	`, remoteName, owner, freshness.Seconds(), lease.Seconds())
@@ -60,14 +75,27 @@ func (db *DB) ClaimGitHubSyncLease(ctx context.Context, remoteName, owner string
 	return true, etag, nil
 }
 
-// ReleaseGitHubSyncLease records the completed scan and frees the lease. Only the
-// owning replica may release; last_synced_at advances so the next poll waits a
-// full freshness window, and etag is persisted for the next conditional request.
-func (db *DB) ReleaseGitHubSyncLease(ctx context.Context, remoteName, owner, etag string, syncedAt time.Time) error {
+// releaseSyncLease records a scan outcome and frees the lease; only the owning
+// replica may release. Success advances last_synced_at, persists the etag and
+// clears any retry. Failure leaves last_synced_at and etag untouched (the last
+// good metadata keeps serving) and schedules next_retry_at with exponential
+// backoff from the consecutive-failure count, never before res.RetryAt.
+func (db *DB) releaseSyncLease(ctx context.Context, table, remoteName, owner string, res provider.SyncResult) error {
+	var retryAt *time.Time
+	if !res.RetryAt.IsZero() {
+		retryAt = &res.RetryAt
+	}
 	_, err := db.Pool.Exec(ctx, `
-		UPDATE github_rpm_sync_state
-		   SET last_synced_at = $3, etag = $4, sync_lease_owner = '', sync_lease_expires = NULL
+		UPDATE `+table+`
+		   SET last_synced_at = CASE WHEN $3 THEN last_synced_at ELSE now() END,
+		       etag           = CASE WHEN $3 THEN etag ELSE $4 END,
+		       sync_failures  = CASE WHEN $3 THEN sync_failures + 1 ELSE 0 END,
+		       next_retry_at  = CASE WHEN $3 THEN GREATEST(
+		                            now() + make_interval(secs => LEAST($5 * power(2, LEAST(sync_failures, 20)), $6)),
+		                            $7::timestamptz)
+		                        END,
+		       sync_lease_owner = '', sync_lease_expires = NULL
 		 WHERE remote_name = $1 AND sync_lease_owner = $2
-	`, remoteName, owner, syncedAt, etag)
+	`, remoteName, owner, res.Failed, res.Etag, res.Backoff.Seconds(), res.MaxBackoff.Seconds(), retryAt)
 	return err
 }