App volumes for DR

[Chris_Nodak]

I have been following this article to rebuild our DR site for our VDI environment and I'm running into challenges with the App Volumes component.

It's recommended to set up an NFS share that then can be mounted at both sites and the data replicated across the WAN. However we cannot get the NFS volume in our production site to mount at the DR site. I've had Broadcom support look at both the server side and network side and they offer no ideas as to why it will timeout and not connect.

My network manager is working with our SD WAN vendor thinking it could be that which is causing the problem as they identified an issue with the appliance in our DR site, but it is slow going waiting on a fix.

What I want to ask is if anyone knows of another way to replicate App Volumes data between the sites without using an NFS share. 

Thanks in advance.

[Robin Harmsen]

Unfortunatly the syncing feature needs/uses a NFS share for the replication.

If i understand correcrly, U are using one NFS share on the primary site and having issues connecting to it though WAN.

What might be possible is to setup aNFS share in your DR location which you keep in sync with some other tooling (I am not aware on how to sync NFS.. might be dependend on the NFS supplier/software)

[StephenWagner7]

Yeah, it sounds like this is an NFS configuration issue, and not directly related with App Volumes. Do you have anyone from your ESXi/vSphere storage team involved? Someone with NFS experience should be able to help you get this going.

[Sean Massey-1]

This almost sounds like a network issue as a “timeout and not connect” could be that the NFS array is unreachable. Another thing to check would be your NFS export configuration - if it’s not configured to allow your DR hosts to connect, you will get a timeout error.

[Ivan de Mes]

I have seen something similar when using NFSv4.

Did you try NFSv3 to rule out the NFS version?

[JamieK]

You can do this using separate NFS shares provisioned from a Windows file server at each site and then use DFS-R to replicate between the two. It doesn't have to be done using a single NFS share presented to both sites. Just be aware that if using Windows to present NFS shares, all the App Volumes VMDKs will be inflated to their thick provisioned sizes based on the templates used to create the packages.

This method is a bit clunky but does work where it is not possible to present a single NFS share across both sites.

[Chris_Nodak]

For some reason I wasn't being notified of replies, I'll have to check my settings. Thanks all for your suggestions. I should have provided more details.

We don't have a storage device that supports NFS. I created an NFS share on a Windows server. It is mounted fine to our hosts in the production site. And the appvolumes data replicated to it from the production datastore.

We've worked with our firewall provider to ensure everything is open and we can see the traffic across the SD WAN but it just times out.

I am using NFS4.

In doing some random testing I had created the same type of share in the DR site and managed to connect 6 of 7 hosts to that share, one timed out. Additional testing, I removed the datastore from one host then it couldn't connect again.

Long story short it's probably something networking/firewall related but we're stumped.

[Graeme Gordon]

Remember that it is a vSphere host that is actually doing the copying of the package files from one datastore to another. You also don't need all the hosts in each site to be able to connect to the shared NFS datastore. To troubleshoot, you may want to reduce this to just one from each site.

[Chris_Nodak]

Thanks for responding Graeme. We have been testing by attempting to connect just one host from the DR site during this process, but it times out every time. I'm not sure where we go from here.