Hi everyone, This is my very first post here and I'm asking because I cannot find the right answer to a problem we have. I have one customer who has a unique server running Linux + MySQL + Apache on the local disk but he wants to improve his systems because of the load (more than 3.000 users simultaneosly accessing the website). The thing is he has bought two servers (both identical) with 24GB RAM each and a local disk to install the OS (saying Debian) plus a storage appliance for centralized storage of the DBs. (this one: http://go.iomega.com/en-us/products.../px4-300r-sc/?partner=4760#tech_specsItem_tab) He basically needs a High Availability cluster for MySQL. How can this be done? I've read about using GFS but I'm not so sure how to make it work with an iSCSI device for HA. I attach a pic with the physical topology (sorry, only MSPAINT technology this time ) If you can help, I would appreciate it.
Wow, surprised no answer? Was just reading up on some clustering stuff and found this thread. It depends on how you want to do this. If you want an active/passive type cluster, then I would OS cluster (CentOS/Redhat or SLES). All things that need to failover is on shared storage. If you need to run active/active, the power of 2 servers masquerading as 1, you will need to cluster MySQL and Apache at the app level. From your diagram, only having two servers, I would run on an OS cluster and run it active/passive with two resource groups, mysql as one active/passive (active on node01) and apache as active/passive (active on node02). That way you are using both servers, and you also have failover for redundancy. Then you need to figure out the resources needed by one node to carry both applications (apache/mysql) for X amount of time if one node does fail. Does that help?
