Failed HA due multisite network subnets

[bugo]

Hi all

I'm in testing phase for multisite (1Gbps link) Windows 2012 Cluster. I have a problem connecting them due different subnets.
To simplify it, my network looks like this:

Site A DC: 10.54.100.0/24 & 10.54.120.0/24
|
1gbps vpn tun
|
Site B DRC: 10.54.110.0/24 & 10.54.130.0/24

As you figured out, i'm not able to pass the network validation step for HA, where it is required to have 2 x identical subnets each site (which cant be accomplished in my case).

So, what is your solution?

Thanks,
Bugo

[jeddyatcc]

I would recommend that you stretch the subnet for the Sync Channel. I know that this can pose some problems, but if you can meet the latency requirements that is probably the only way to make it work. If you are using the 3 node setup, I think you can use different subnets (haven't tested, but that "should" be the way it works) for the 3rd replica node.

[bugo]

I would recommend that you stretch the subnet for the Sync Channel. I know that this can pose some problems, but if you can meet the latency requirements that is probably the only way to make it work. If you are using the 3 node setup, I think you can use different subnets (haven't tested, but that "should" be the way it works) for the 3rd replica node.

Thanks for reply. Unfortunately it's not a restriction by latency or stretched subnet, but due simply fact that starwind validates the network only by a simple match (plain text if i may say so). I did a dumb test by adding to each node a secondary ip (from the other side) to one of the eth and now it looks like starwind recommended config:

Site A DC (Host 1): eth1 - 10.54.100.101 (real), eth2 - 10.54.120.101 (real), 10.54.110.102 (no access to site B where this subnet is used)
|
1gbps vpn tun
|
Site B DRC (Host 2): eth1 - 10.54.110.101 (real), eth2 - 10.54.130.101 (real), 10.54.100.102 (no access to site A)

Really need to figure this out! I would appreciate any help.

Thanks

[Anatoly (staff)]

Its all actually been said already - you need to have IPs in same subnet. Before we will move further - may I ask you why having same subnets is not an option for you?

[bugo]

Its all actually been said already - you need to have IPs in same subnet.

Hi Anatoly. Sorry, i've read all materials and searched the forum, but didn;t find anything. Dont get me wrong, but for my self knowledge, i would appreciate any link to clarify why this cant work on different subnets (it's 1st app that i;ve encountered with this limitation). Following your advice...

Before we will move further - may I ask you why having same subnets is not an option for you?

those 2 sites are connected by a tunnel, without a L2 bridge (necessary to run same subnet on sides) and since networks are not my strong skill, i've asked for help to my network admins and they will try (have ) to implement it asap.

I'll keep you posted. Thanks again

Bugo

[jeddyatcc]

In speaking to my network guys. They said that in order to stretch a subnet in our office across an internet vpn tunnel would require static routes to be create for each IP Address, or we would have to split the subnet into smaller chunks. I believe that this makes a multi-site "tunneled" system very difficult. Can anyone confirm whether the 3rd async replica can be placed in a different subnet? If not I need to rework my current project for DR.

[Anatoly (staff)]

bugo, had you any chance to re-configure the system as you have mentioned?

those 2 sites are connected by a tunnel, without a L2 bridge (necessary to run same subnet on sides) and since networks are not my strong skill, i've asked for help to my network admins and they will try (have ) to implement it asap.

Thank you

[bugo]

Hi Anatoly,
Bridge is up & running and successfully configured starwind. I'm currently setup the win cluster, so hope will work smoothly.

You can close this post topic.

Thanks again for your support.

[anton (staff)]

Thank you very much for keeping us updated!