This feature was earlier a port-in-progress in the OurDelta builds, but wasn't included in the work for 5.1 that Monty and I did for MariaDB.
It's in stock MySQL 5.6. It's a useful feature, and clients still sometimes need it. The pt-slave-delay tool does the job, but it's nicer to have it in the server - and heck, the code already exists.
If we can get this merged into (for instance) 10.1, that'd be great.
thanks
Would love this feature - I'm struggling with pt-slave-delay issues where it is behaving erratically, and obviously you have to run it as a separate process. Plus, there are quirks with it such as that it will die after 10 failed attempts to connect, and it's not 'guaranteed' to keep things delayed as much as if it was built into the database.
Phil Sweeney
added a comment - Would love this feature - I'm struggling with pt-slave-delay issues where it is behaving erratically, and obviously you have to run it as a separate process. Plus, there are quirks with it such as that it will die after 10 failed attempts to connect, and it's not 'guaranteed' to keep things delayed as much as if it was built into the database.
Waiting for this feature too. It can be useful in a testing environment to detect failures on the client side because of asynchronous replication delay (which may occur on a master/slave production environment).
Nicolas Payart
added a comment - Waiting for this feature too. It can be useful in a testing environment to detect failures on the client side because of asynchronous replication delay (which may occur on a master/slave production environment).
Waiting for this feature, and hope it in the 10.1.x series. It will help the users who use the replication as a buffer for preventing the wrong sql usage.
Su, Jun-Ming
added a comment - Waiting for this feature, and hope it in the 10.1.x series. It will help the users who use the replication as a buffer for preventing the wrong sql usage.
Waiting for this as well. Right now we're using pt-slave-delay which works but isn't stable; our sysadmin is yelling at me to switch to Percona. I would prefer to stay with MariaDB but data integrity trumps all other concerns.
Simon Wright
added a comment - Waiting for this as well. Right now we're using pt-slave-delay which works but isn't stable; our sysadmin is yelling at me to switch to Percona. I would prefer to stay with MariaDB but data integrity trumps all other concerns.
Phil Sweeney
added a comment - For those tracking this bug, knielsen has begun work on this and posted an update on the developers mailing list:
https://lists.launchpad.net/maria-developers/msg09993.html
Thanks Kristian!
Task
Critical
Would love this feature - I'm struggling with pt-slave-delay issues where it is behaving erratically, and obviously you have to run it as a separate process. Plus, there are quirks with it such as that it will die after 10 failed attempts to connect, and it's not 'guaranteed' to keep things delayed as much as if it was built into the database.