Assuming one starts with one server as master and then adds a few more slaves to the cluster, what happens if the master goes down, does the whole cluster go down (the idea being that clustering makes for stability)?
How can one upgrade the master (designate a slave to be master or is transfer of master backup to the new master necessary), will there be cluster-down-time?
Apologies for delay. I just noticed you posted 2 new posts
IW did create a cluster failover when Centos 6 was current. However, this was never available for sale and I can tell you it worked lovely (I was one of the beta testers)
So, you can add as many slaves to the master, which increases the availability of services that you can handle at any one time but if teh master goes down, all slaves will fail as well
However, that said, there is a few ways around this
create a duplicate master and rsync from master to master and if master goes down, change IP and add slaves in
have /home on a NFS and if master does down, you can quickly create a new master and add slaves in
create a new master and restore /home etc… then add slaves in
all the above are quick to do (circa around 1 Hour I think depending upon sizes of data)
Thank you so much, really appreciate it!
If that failover solution was not for sale, is it anywhere available for free/beta or could one convince the IW wizards to update it and include it or make it available for extra charge?
Your manual solution sounds doable.
I wonder, to avoid even a short disaster, should one put the master perhaps on a nearly unfailable cloud like one of the big cloud providers or what underlying infrastructure would you suggest?