For IV 5.4+, this retry can be done by a new feature called GearmanMoveWorker. This MoveWorker doesn't need to login to the database or restart any services, but move all items in _fail queues to normal queues.

When a gearman worker process encounters a job it cannot complete it will retry a configurable number of times, and then put the item into a different gearman queue which is named *_fail (e.g. jobs from the vad queue are placed in vad_fail).  If the failure was caused by an issue which is now resolved, you can return the jobs to the original gearman queue by modifying the database as described below.

Before you start

You will need a database user account with permissions to modify the gearman.gearman_queue table.  You can find the database root user password by looking for mysql_root_password in the install.cnf file you used when you installed IV.

This article uses SQL commands to modify the database.  You are advised to backup the database before starting.  You can use any MySQL database tool for this task or the "mysqldump" command line tool on the IV server.   An example command to back up the database to /var/backups is:

/usr/bin/mysqldump -uroot -p --all-databases --single-transaction | gzip > /var/backups/full_backup_`date -Iminutes`.sql.gz

Retrying jobs

To retry jobs that failed in Tagger you should move them from the tagger_fail queue to the tagger queue.  The MySQL query is:

update gearman.gearman_queue set function_name = 'tagger' where function_name = 'tagger_fail'

To run this using the mysql command line client as the root user:

sudo mysql -uroot -p -e "update gearman.gearman_queue set function_name = 'tagger' where function_name = 'tagger_fail';"

Then restart the gearman service to reload the jobs from the database:

sudo systemctl restart gearman-job-server