I try to set up a (pretty) high availability solution for a service that I run on Linux machines and here's what I come up with:
- one front end server used for balancing. IT should have haproxy installed on it and direct the requests to 3 other servers
- 3 main servers that will use for
MySQL - galera so one server can go down while the others still be sync and answer queries
gluster - for replicating the files between these 3 servers
nginx - will just have pretty much the same configurations on all the servers so it will be independent.
I was wondering if you have any possible suggestions for maybe improving this solution or maybe you have encountered different pitfalls while configuring/using these services that you'd like to share with me.