The main feature of this tool is that it has two disk usage thresholds which determine when it purges your binary logs:
- a lower, "nominal", threshold above which binary logs will be purged if-and-only-if none of the replication slaves are still reading it,
- and a higher, "critical", threshold at which the behavior is configurable.
For those not familiar with the MyCAT project, it is an open-source Perl toolset for managing MySQL/Linux servers that I wrote initially for my own use, but after a few people asked for similar tools I decided to publish it. Currently, it is composed of three programs that: monitor replication (rep_mon), monitor and rotate binary logs (binlog_mon), and allow remote shell and scp access to predefined groups of servers (rcall). This third tool is great for simplifying tasks such as syncing a new httpd.conf file and running a rolling-restart across a large web farm. All three tools read the same XML configuration file which defines servers and their properties (such as which servers have mysql, what volume the binlogs are on and what the nominal/critical levels are, what an acceptable replication "lag" is, and whether or not exceeding that lag is important enough to send you an email, etc). It really does a lot of things useful to any linux or mysql admin working in a LAMP environment, and I encourage anyone to send me feedback or feature requests.
MyCAT on Sourceforge