dev-db/mysql-init-scripts-2.2 was confirmed working, I worked on it with Brian Evans. 2.2-r1 cannot form a cluster. I don't know all of the issues yet, the *obvious* error is that the 'bootstrap' action was changed to 'bootstrap_galera' but mispelled as 'boostrap_galera'. If I fix that, the first node can be bootstrapped, but subsequent nodes fail to join the cluster. I manually rolled back to v2.2 and was immediately able to fully start my cluster again. Some change in 2.2-r1 has broken the cluster startup logic. I haven't had the opportunity to compare the two and study the issue in depth yet.
Besides the typo, very little changed between 2.2 and 2.2-r1... https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=d97f1c09eb25b28a65aedafa11a0c850a6438c7d Nothing has changed to affect other nodes. Typo fixed in 2.2-r2
Huh. I wonder why cluster formation failed then....? I'll retest with 2.2-r2. Maybe the problem is actually with MariaDB Galera Cluster rather than the init scripts themselves. I was starting from all nodes down after a power outage, *but* all nodes did try to do a 'start' before I had the opportunity to bootstrap a first node. It's possible that screwed something up. Honestly I don't think MariaDB did nearly as clean and robust a job of the Galera integration as Percona did.
> Maybe the problem is actually with MariaDB Galera Cluster rather than the > init scripts themselves. I was starting from all nodes down after a power > outage, I should note, all nodes CLEANLY down — I had time to bring everything down cleanly before UPS power ran out. Perhaps what I need to do is write a tool to detect the condition "all servers are up but no node is started" and pick a first node to bootstrap.
