Comments on: MySQL replication monitoring with Python

By: Greg

Greg — Thu, 15 Dec 2011 18:58:15 +0000

Honestly, I’m not an expert on what you’re doing here. I am doing something much more basic – digging into the status of a slave and sending alerts with monit if there is a problem.

I am fairly new to python, though and your code was a very helpful basis.

My comment was only to help for someone who used your code but was using a version of mysql that would break on the seconds_behind_master check. (also, it is a simple fix). It will of course, break if someone changes the name of that variable in a future release.

By: tom

tom — Thu, 15 Dec 2011 17:52:01 +0000

Greg, Thanks!

Sorry to say I haven’t re-visited this subject in a while (posted this 2.5 years ago).

I actually had some qualms regarding the whole setup and its robustness. What is your opinion of the method / algorithm as a whole?

By: Greg

Greg — Thu, 15 Dec 2011 17:35:54 +0000

You can get the results from “show slave status” to be in a dict form by using a DictCursor. This lets you specifically look at ‘seconds_behind_master’ instead of just assuming it is the last thing in your row (which it isn’t in 5.5).

cursor = conn.cursor(MySQLdb.cursors.DictCursor)

By: tom

tom — Tue, 01 Mar 2011 12:44:28 +0000

That is definitely a viable method. The solution suggested here is not by any means the optimal solution.
In addition this is sort of an end case – not the usual way of handling mission critical data with MySQL replications.
Thanks!

By: Danial Poeschel

Danial Poeschel — Tue, 19 Oct 2010 21:08:48 +0000

I depend on having each model backed up and watching each appliance, as opposed to simply monitoring naturally on the Internet, after which responding in a short enough point in time for a resolution.