This week started with some heavy testing for setup_replication.sh, and some bugs did come up.
The new version should work in more systems than before, and should be more stable. The bottom line: the chances to end up with replication working are greater now :)
Work is now focused on improving the core cluster code, with several bugs being ironed out from the slave routine and the takeover procedure. The rc-script has also been fixed, and even though it's 'design' is not something I'm proud off (I guess it's good to know I've learned a lot about writing init scripts in the past 4 years, hehe), it does work for Red Hat/Fedora in all scenarios now.
What will come next, in the short term:
- use of the ssh agent in order to allow the use of passphrases for key based ssh
- uso of sudo to run anything that requires root (service mysqld restart, ifconfig, fake, etc.)
- Heavy testing under CentOS on our side, with a RHEL4 system that's being offered by a user. Let's hope good things come out of this too.
