After 7.2.1 update action scheduler and debug log have gone crazy
-
Action scheduler actions and logs don’t clear and get millions of rows in the DB. It produces about 2MB of data a minute.
Debug log increases about 1GB per 20 mins with millions of these errors.
[21-Dec-2022 09:40:09 UTC] WordPress database error Unknown column ‘wp_postmeta.post_id’ in ‘where clause’ for query SELECT meta_id FROM m31w_postmeta,
(SELECT DISTINCT post_id FROM m31w_postmeta
WHERE (meta_key = ‘_billing_country’ OR meta_key=’_shipping_country’) AND meta_value=’NZ’)
AS states_in_country
WHERE (meta_key=’_billing_state’ OR meta_key=’_shipping_state’)
AND meta_value=’NL’
AND wp_postmeta.post_id = states_in_country.post_id
LIMIT 100 made by do_action_ref_array(‘action_scheduler_run_queue’), WP_Hook->do_action, WP_Hook->apply_filters, ActionScheduler_QueueRunner->run, ActionScheduler_QueueRunner->do_batch, ActionScheduler_Abstract_QueueRunner->process_action, ActionScheduler_Action->execute, do_action_ref_array(‘woocommerce_run_update_callback’), WP_Hook->do_action, WP_Hook->apply_filters, WC_Install::run_update_callback, wc_update_721_adjust_new_zealand_states, Automattic\WooCommerce\Database\Migrations\MigrationHelper::migrate_country_states, Automattic\WooCommerce\Database\Migrations\MigrationHelper::migrate_country_states_for_orders
-
Same thing happened to me this morning. After deleting a 26 GB Apache error log file to free up some disk space, I managed to transfer the previous day’s (merely 12 GB!) Apache error log file to my local system to investigate, and I discovered it was full of millions (seriously) of lines like this:
[Wed Dec 21 00:00:26.821990 2022] [php:notice] [pid 237087] [client REDACTED:47642] WordPress database error Unknown column 'wp_postmeta.post_id' in 'where clause' for query SELECT meta_id FROM REDACTED_wp_postmeta,\n\t\t\t\t\t(SELECT DISTINCT post_id FROM REDACTED_wp_postmeta\n\t\t\t\t\tWHERE (meta_key = '_billing_country' OR meta_key='_shipping_country') AND meta_value='NZ')\n\t\t\t\t\tAS states_in_country\n\t\t\t\tWHERE (meta_key='_billing_state' OR meta_key='_shipping_state')\n\t\t\t\tAND meta_value='SL'\n\t\t\t\tAND wp_postmeta.post_id = states_in_country.post_id\n\t\t\t\tLIMIT 100 made by do_action_ref_array('action_scheduler_run_queue'), WP_Hook->do_action, WP_Hook->apply_filters, ActionScheduler_QueueRunner->run, ActionScheduler_QueueRunner->do_batch, ActionScheduler_Abstract_QueueRunner->process_action, ActionScheduler_Action->execute, do_action_ref_array('woocommerce_run_update_callback'), WP_Hook->do_action, WP_Hook->apply_filters, WC_Install::run_update_callback, wc_update_721_adjust_new_zealand_states, Automattic\WooCommerce\Database\Migrations\MigrationHelper::migrate_country_states, Automattic\WooCommerce\Database\Migrations\MigrationHelper::migrate_country_states_for_orders, referer: https://REDACTED/wp-cron.php?doing_wp_cron=1671580796.8379271030426025390625I have redacted my site name and WP database table prefix here, but this is clearly the issue. There’s a SQL query in something ActionScheduler is doing that is hardcoded to the
wp_postmetatable, rather than using the site’s table prefix variable.Oops.
@chrisplaneta – Would you please be able to elaborate what is needed in point 4: “find and enable some “maintenance mode” plugin“. Is it a specific plugin you need to install? Or, do you have an example of such a plugin?
UPDATE: I believe I get you. You need a plugin that puts your site into maintenance mode, for when you restart every plugin.
@chrisplaneta – I followed your solution guide,
and luckily it worked well on my side as well. Thank you!and first it looked like it was working. However, thewpcc_actionscheduler_actionsandwpcc_actionscheduler_logstables, plus the.php.error.logfile started filling up again. It seems like I need to wait for 7.2.2.UPDATE:
I followed the above mentioned procedure, without re-activating WooCommerce. Then, the log-o-maniac stopped, and this site still works fine. Looking forward to version 7.2.2, where this should be fixed completely.Wondering about clearing out the ActionScheduler data tables (mine have 2.6 million entries after this… situation).
For those of us comfortable with direct database actions, is it fine to just truncate the
wp_actionscheduler_actionsandwp_actionscheduler_logs(prefixes removed) tables? I’d rather not install a random database cleanup plugin I’ve never heard of before to do something this simple, if that’s all there is to it.(Update: OK, well… I just decided to go ahead and do it after doing a mysqldump of the database as a precaution.)
- The topic ‘After 7.2.1 update action scheduler and debug log have gone crazy’ is closed to new replies.
