example storage cleanup diagnostics

Observe stored-file cleanup failures

Experimental. A failed safety check stops the remaining daily cleanup sweep. Individual unlink failures are reported while later candidates continue. The next cron run retries retained bytes; this action is for diagnostics, not for forcing a retry or bypassing the checks. No SQL or filesystem paths are exposed.

<?php
/** Plugin Name: OpenStation Cleanup Diagnostics */
defined( 'ABSPATH' ) || exit;

add_action( 'openstation_stored_files_reconcile_failed', function ( $error ) {
    $stage = $error->get_error_data()['stage'] ?? 'unknown';
    // Forward a small, non-sensitive event to your monitoring integration.
    do_action( 'my_monitoring_event', 'openstation-cleanup-failed', array(
        'code'  => $error->get_error_code(),
        'stage' => $stage,
    ) );
} );

Stages are orphan_rows (candidate scan), delete_row (lock, revalidation, or conditional delete), known_bytes (registered names), and delete_bytes (lock or revalidation). unlink_row_bytes and unlink_bytes report filesystem failures without aborting the sweep. SQLite reports a lock-stage failure because its advisory-lock shim cannot safely coordinate destructive cleanup. The code is openstation_reconcile_failed. Healthy files with active or trashed placements are retained.