Here are the things I can't change: We got a mySQL myISAM replication running a master and some slaves. One slave is being used for nightly backups using mysqldump. The backups is using heavy locks forcing it's slave to "age" compared to the master. mysqldump is connecting as root dumping all the databases while every other connection is using one non-SUPER-user per database. The dumps are started using a simple shell script on a dedicated backup server.

Here is the problem: I'ld like to detect if a backup is running or not while being connected as a "normal" user to one of the databases running on the server being backuped. Cronjobs should be able to react on the running backup by exiting early or processing only really important required jobs which can't wait until the backup is done. The server running the backup script has no connections to the webservers, their only common point is the mysql server. The backup script could do some things to flag the server as "currently running a backup". I'd prefer to set some kind of "global server-side environment variable visible by all users" to ignore only the one slave running the backup but it would also be ok to simply know that a backup is running on that database or replication chain. Using the database for this is no option as it's being locked during the backup.

I tried to create a temp table on backup start (existing = backup is running, not existing = no backup running), but temp tables only show up for the connection which originally created it (as expected). I googled if a non-SUPER-user could get the permission to see the SHOW PROCESSLIST of all connections including other users and root (any connection from the backup server = backup is running) - but no luck. Giving SUPER to all database users is no option.

Any suggestions?

Thanks for reading (and answering).

migrated from stackoverflow.com Jan 13 '12 at 9:37

This question came from our site for professional and enthusiast programmers.

  • Do you have shell access? Does the backup run locally or from a remote host? Would pgrep mysqldump be an unacceptable choice or does it have to be from within MySQL? – Aaron Brown Jan 14 '12 at 16:37
  • Yes I have shell access, but the backup is being created by a dedicated server and SSH'ing from running (web-)scripts to the backup server only to see if it's running something would be too risky. – Sebastian Jan 16 '12 at 7:15

Your only hope in this scenario is to have the PROCESS privilege.

The SUPER privilege allows you to kill processes. Naturally, you don't want that. On the other hand, the PROCESS privilege allows you to see the processlist.

According to the MySQL Documentation on the PROCESS Privilege

The PROCESS privilege pertains to display of information about the threads executing within the server (that is, information about the statements being executed by sessions). The privilege enables use of SHOW PROCESSLIST or mysqladmin processlist to see threads belonging to other accounts; you can always see your own threads.

OK, Big Deal. The PROCESS Privilege lets you see the Process List. How does that help?

You can quickly detect a mysqldump in progress when you run SHOW PROCESSLIST; and see a pattern like this in the Info field:


If you have the PROCESS Privilege and are running MySQL 5.1+, you can run this query:

select COUNT(1) mysqldumpThreads
from information_schema.processlist
where info like 'SELECT /*!40001 SQL_NO_CACHE */%';

Here is a sample output:

mysql> select COUNT(1) mysqldumpThreads
    -> from information_schema.processlist
    -> where info like 'SELECT /*!40001 SQL_NO_CACHE */%';
| mysqldumpThreads |
|                1 |
1 row in set (0.00 sec)

All you need to do is query for that number. If you get a nonzero, mysqldump must be running.

Give it a Try !!!

  • Is that 40001 in the query comment going to remain stable for future versions of mysql? – mc0e Sep 20 '17 at 17:42
  • @mc0e There are only 2 instances where this will become a problem: 1) If you are using mysql 3.23, 2) if this version metatag mechanism is deprecated. Other than that, this will be remain stable. – RolandoMySQLDBA Sep 20 '17 at 17:50

Make your backup script set up a value in some separate table before backup starts, and reset it when it ends. You can then just query for this value.

  • mySQL locks whole databases, no chance to access a table while the backup is running. – Sebastian Jan 13 '12 at 8:57
  • I claim that statement above is not true. Besides, you can use a table that is in database that is not being backed up – Mchl Jan 13 '12 at 9:00

You might want to read up on using LVM snapshots to backup you myisam tables and minimize replication down time to a couple minutes:


If you're able to migrate to innodb you could user xtrabackup and take live backups with next to no noticeable interference.


Personally we were using LVM snapshots for a ~220 GB instance spanning 3 databases. The overhead of the LVM snapshot and load on the disks caused replication to fall behind as much as 8,000 seconds nightly. There wasn't any locking but the poor little disks just couldn't keep up with the heavy replication load and the backup. We had to switch to fusion io. That's certainly nice -- but very costly.

  • 1
    Just as a reminder: xtraBackup's point-in-time snapshot is based on the completion time of the backup rather than the start time. – RolandoMySQLDBA Jan 13 '12 at 17:22

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.