This is the basic plan:
- The latest
PhpMyAdminwill be installed on the cluster, removing that dependency on Lettuce. This should happen soon.
- Chard will be rebuilt to take on two tasks: NFS (replacing Arugula), and
Subversion(replacing Lettuce in this role). This can happen basically any time, with only brief downtime for
Subversion. We'll end up with a more suitable machine for the NFS role.
- Arugula will be rebuilt for two roles: first, as another cluster node for webserver.hcmc.uvic.ca, and second, as a backup db server in case something happens to Cress. This gives us cover for Cress, but meanwhile the machine is doing something useful. This can happen any time.
eXistinstances need to be run on Pear, where each will run on a different port. The code itself lives in home1t, which Pear mounts, so that shouldn't be complicated, but each project is slightly different (MOL runs the full Cocoon instance, whereas FT just runs the DB). We need to decide the ports first; RE will open up those ports to the cluster in the ACL. Then we shut down the instances on Lettuce, and start them up again from Pear. Once we confirm that both projects run OK, Lettuce will be cleared for a rebuild.
- Lettuce will be rebuilt as a third cluster node. We're tentatively scheduling this to start in the second week in January.
- Access to projects on the cluster will be through
webserver.hcmc.uvic.ca, but we will also have
arugula.hcmc.uvic.caaccess so that we can test and debug issues we might suspect with specific machines.
- Our disk space will probably be reconfigured at some point, so that half (2TB) is live, and half (2TB) is archive. This is isn't scheduled yet, and perhaps should wait until all the other changes have gone through.
- Java is not on the cluster machines, and we have no reason to install it on them for the moment.
- A virtual machine will be created as backup for Pear.