MySQL 5.6 GTIDs replication and HAProxy

I've been thinking about this for a few weeks now.  We currently run a Percona XtraDB cluster with HAProxy installed on each of our application servers.  While this provides some good redundancies, the HAProxies don't know how busy the DB nodes are and it won't continue to scale well.

To that end, we're looking to break up the Percona XtraDB cluster into a more traditional Master - Master setup with multiple slaves under each master and HAProxy installed on a pair of servers with a VIP handled by CARP (technically ucarp).

But I'm looking towards the future of MySQL 5.6, the new GTIDs replication, and how to setup HAProxy for that configuration. To that end, in the next week or so I'm going to setup some virtual machines to test this all out with.

Copying one file to multiple directories

Just a simple post to remind me how to do this in the future.

I needed to copy a new image to sixty (60) client directories, and symlinking like we would normally do was not an option. Thankfully xargs came to the rescue once again.  I found my answer here, but I wasn't going to type out the full paths for each of the sixty directories.

# echo /somedir/*/pub/product/images/ | xargs -n 1 cp -va preview_unavailable.png

FreeBSD Crontabs location

This is a quick blog post, mainly for my notes.  In FreeBSD the users crontabs are stored in /var/cron/tabs by default. This is different than I'm used to coming from RHEL and CentOS world where crontabs where stored in /var/spool/cron/

FreeBSD Cron documentation: http://www.freebsd.org/doc/handbook/configtuning-cron.html

FreeBSD sed -i Error "sed: 1: "FILE": bad flag in substitute command: '.' "

On FreeBSD if you use the "-i" flag with sed you must specify a backup file.  The simple fix, if you don't want a backup file is to use "" (two quotes, no space between them).  You could also use any other filename you wished.

Examples:
BAD: sed -i s/192.168.1.1/192.168.1.44/g file.txt
GOOD: sed -i "" s/192.168.1.1/192.168.1.44/g file.txt