this is not meant as a proper PR, but rather as a way of showing you the changes I needed to make for me to be able to get the MariaDB Galera cluster up and running on my Kube 1.7.x cluster, based on your repo here and your blog article.
Thanks a lot for that, they helped a lot in getting started.
Below is the commit message that explains some of the changes. It's actually a bit inaccurate though (I'm noticing only now; have squashed lots of commits), since I've actually done a full rewrite of the healthcheck.sh script based on the Galera Cluster spec and what the docs say how to tell the current node status.
- remove HEALTHCHECK instruction from main Dockerfile as that conflicts
with Kubernetes health and liveness probes; added a swarm.Dockerfile
that still contains that instruction
- fix XTRABACKUP_PASSWORD handling from env to my.cnf
- optimized healthcheck.sh to work with local mysql instance and connect
through socket, because that's what we're dealing with here
- also fixed minor formatting issues in that script
This is not in production for me yet, but it's soon going to be.
Hi there,
this is not meant as a proper PR, but rather as a way of showing you the changes I needed to make for me to be able to get the MariaDB Galera cluster up and running on my Kube 1.7.x cluster, based on your repo here and your blog article.
Thanks a lot for that, they helped a lot in getting started.
Below is the commit message that explains some of the changes. It's actually a bit inaccurate though (I'm noticing only now; have squashed lots of commits), since I've actually done a full rewrite of the healthcheck.sh script based on the Galera Cluster spec and what the docs say how to tell the current node status.
This is not in production for me yet, but it's soon going to be.
Cheers, Denis