Description

The issue encountered was that when running cron via linux cli using a command like this

/var/www/api.inkjin.com/vendor/drush/drush/drush --root=/var/www/example.com cron

The value simple_access_log.settings -> delete_log_after was an old, stale value cached in Redis, not the up to date one. The read value was 4 months in seconds, while the correct value was 0.

This caused simple_access_log to delete old logs even though it was configured not to.

The values read by

  • running drush in CLI with the command above
  • running website code
  • using the redis command below
    • redis-cli --raw HGET $PREFIX:config:simple_access_log.settings data

were all correct. Only linux cron was getting a bad value.

The issue turned out to be caused by linux cron being run using a different php command (THE VERSION WAS IDENTICAL: 8.1.32).

Setting the cron to run this command instead, fixes the issue completely:

php81 /var/www/api.inkjin.com/vendor/drush/drush/drush.php --root=/var/www/example.com cron

Root cause unknown.

Steps
  1. use a version-specific php command, and run "drush.php" file instead of "drush"
Notes

  • Redis compression was disabled by removing this config from Drupal settings "$settings['redis_compress_length']")

  • Redis caches were being updated, but somehow, a different value was being returned by Redis when running using the default "php" command as opposed to "php81".

Tags

Recently posted