Queue and delivery

Flow

Tables: modx_indexnow_queue, modx_indexnow_history (with your table prefix).

Queue tick (primary background)

The plugin listens to OnWebPageComplete (front end) and OnManagerPageAfterRender (manager). With IndexNow enabled and indexnow_queue_enabled = Yes, after enqueue the code calls scheduleQueueTick(): once per request, via register_shutdown_function.

One tick:

• takes a MODX cache lock (indexnow_queue_tick, TTL 55 seconds) so parallel requests do not run duplicate workers;
• processes at most 25 due URLs (QUEUE_TICK_BATCH_MAX), even if indexnow_batch_size is higher;
• if Scheduler is installed, calls ensureScheduledRun() as backup cron.

Scheduler is not required for background delivery. Tick covers the usual case after a resource save or a manager page load.

Plugin events

Event	Behavior
OnDocFormSave	Published resource → update. Unpublished → delete.
OnResourcePublish	Same as saving published → update.
OnResourceUnPublish	→ delete.
OnBeforeDocFormDelete	Stores URL before delete.
OnDocFormDelete	Enqueues delete for the stored URL.
OnWebPageComplete	Schedules queue tick after the front-end response.
OnManagerPageAfterRender	Schedules queue tick after the manager response.

Update enqueue applies to published, non-deleted resources with a resolvable absolute URL. Hosts localhost, private/reserved IPs, and metadata.google.internal fail URL validation.

If publishedon is in the future, the row waits with available_at = publishedon.

Deduplication

Open rows (pending / processing) are unique by host + url.

Saving the same page again does not create duplicates. action, available_at, and metadata are updated.

Last event wins. Example: unpublish sets delete, republish rewrites the same row to update.

Worker

One pass:

1. Reset stale processing older than 15 minutes to pending.
2. Claim pending rows with available_at <= now, limited by indexnow_batch_size (or less on tick).
3. Group by host.
4. POST batch(es) to the endpoint per host group.
5. Write history and update the queue.

If the key is invalid or indexnow_endpoint fails validation, the worker exits without sending (message in the MODX log). The queue grows with no history rows.

HTTP codes

Code	Behavior
200, 202	Success. Row leaves the queue; history status success.
429, 5xx, network / timeout	Temporary. Retry after indexnow_retry_delay until indexnow_max_attempts.
400, 403, 405, 422	Permanent. Status failed; no endless automatic retry.
Other 4xx (e.g. 401, 404, 410)	Also immediate failed, no retry series (not treated as temporary).

A successful IndexNow response means the notification was accepted, not that the page is already in search results. Same in Yandex documentation.

IndexNow retry vs Scheduler retry

• Queue attempts / available_at: delivery to the endpoint.
• Scheduler task retry: separate; only when the task runner itself fails.

Manual processing

• Process queue on Status: immediate full worker pass with indexnow_batch_size limit.
• Queue tick and Scheduler pick up due URLs without UI action.