No LIVE data visiblie when replicating

arminb · Mon Jul 14, 2014 2:07 pm

Hello,

I'd like to create a file replication between two servers. Both servers are connected to a router which bonds them. That means they have the same IP address. If a client accesses the IP address the router decides which server responds. If one of the servers fail the router decides to access the other one. For this configuration I need a live data replication. What I got working for now is a replication without live file changes. The changes I make on one server are not visible on the other server until i remount the iSCSI drive.

I found this "issue" on your FAQ: http://www.starwindsoftware.com/starwind-faq#q_12

I hope you can give me a more detailed and concrete hint how to achieve this.

Mon Jul 14, 2014 2:13 pm

Configure a fault tolerant block back end with a pair of StarWind-running services and put a failover SMB3 / NFSv41 file shares on top of it. That would be the right way to go. There's a bunch of a technical papers on our site on how to do this. Link you gave has nothing to do with your design.

Mon Jul 14, 2014 2:30 pm

Here are links:

http://www.starwindsoftware.com/ns-conf ... or-smb-nas

http://www.starwindsoftware.com/ns-conf ... or-nfs-nas

Hope this helped

arminb · Mon Jul 14, 2014 2:48 pm

I found these sheets already. This configuration requries a (central) domain controller which I don't (want to) have. I'd like the servers to be peer accessed. The need of a domain controller creates a single point of failure which destroys the redundance aspect

Mon Jul 14, 2014 4:23 pm

It does not create any SPOF. Typically you have multiple DCs. Some of them running inside a VMs (that what you can do no problem) with a physical one (optional) used as a backup. See for example:

http://community.spiceworks.com/topic/2 ... c-failover

http://community.spiceworks.com/topic/5 ... nvironment

Hope this helped

arminb wrote:I found these sheets already. This configuration requries a (central) domain controller which I don't (want to) have. I'd like the servers to be peer accessed. The need of a domain controller creates a single point of failure which destroys the redundance aspect