How it works

• Part of: LH Wayback Machine

Each site keeps its own queue. Nothing is stored outside post and term meta, and no tables are added.

Lifecycle of an item

1. Queued. A published post or public term that has never been sent, or has changed since it was last sent.
2. Sent. Once the item has been unchanged for a minute, is not open in the block editor, and was not sent in the last 15 minutes, a Save Page Now request is fired without waiting for a reply.
3. Checked. An hour later the Wayback Machine CDX index is queried for a successful capture newer than the send. Checks repeat hourly.
4. Archived when a capture is found; the column links straight to it. After two hourly misses the item is resubmitted, this time waiting up to 30 seconds for a reply. After two resubmits without a capture it is marked Failed (typically pages behind a login, or ones the archive refuses).

Editing an item at any stage puts it back to Queued and starts the cycle again. Its last known snapshot stays linked until a newer one is confirmed.

Stored data

• _lh_wayback_machine-timestamp: when the item was last sent (UTC).
• _lh_wayback_machine-snapshot: the latest confirmed capture, as a 14-digit Wayback timestamp.
• _lh_wayback_machine-checked, -misses, -attempts: verification state.
• _lh_wayback_machine-dirty: terms only, when the term was last edited (terms have no modified date).

Items sent before 2.0.0 are treated as legacy: they get one CDX lookup to record their latest snapshot and are never resubmitted unless edited again.

Background jobs

Jobs run through Action Scheduler when it is loaded and WP-Cron otherwise. lh_wayback_machine_single processes the queue and reschedules itself for the next item that falls due. lh_wayback_machine_wakeup runs daily as a backstop. A per-site lock stops two passes from running at once. Both jobs are removed on deactivation.

Filters

• lh_wayback_machine_init_plugin: return false to disable the plugin without deactivating it.
• lh_wayback_machine_get_applicable_post_types, _get_applicable_post_statuses, _get_applicable_taxonomies: what gets archived.
• lh_wayback_machine_return_settle_delay (60s), _return_reping_gap (900s), _return_check_interval (3600s), _return_max_misses (2), _return_max_attempts (2), _return_batch_size (5): timing and limits.
• lh_wayback_machine_use_action_scheduler, _return_group: queue backend.