I went through a vSphere Replication and Site Recovery Manager implementation this week and overall it went very well. There are a few things that aren't really noted in the documentation that I wanted to get out there just in case anyone else runs into the speed bumps I did this week. While I've implemented SRM with array based replication many times, this was my first go with vSphere Replication.
First, and most important, is to use the FQDN throughout the entire implementation of SRM, VRMS, and VRS. While we used the FQDN for the vCenter server when we installed SRM, when you deploy the vSphere Replication Manager Server (VMRS) it defaults to the IP address for the vCenter server on the configuration page. If you don't change this to the FQDN of the vCenter server SRM won't be able to "find" the VRMS servers when you attempt to create the connection between the two.
SQL Express port issue. This customer decided they wanted to use the vCenter implementation of SQL Express for the VRMS database. While this works fine you need to understand that when vCenter does it's install of SQL Express it chooses a variable port by default. Since it does that, you'll need to manually change the port the DB connections are received on. I recommend port 1433 which is the default SQL Server port. You can follow a good blog post on MSDN here: Link
VRMS and VRS SSH root access. By default root access is not enabled for remote SSH management. While a good security practice, it makes troubleshooting a PitA. There is a great KB article you can read here, but I've listed the steps below if you would rather stay here.
- Open the console for the appliance in vCenter
- login with root. You set this password during deployment of the appliance
- cd to /etc/ssh/
- vi sshd_config
- find line with “PermitRootLogin” and set to yes. (type i to insert and then ESC when done editing. :wq will save changed and quit.)
- restart the SSH service with service sshd restart
- Create a one to one datastore "mirror" so that you can automatically map the datastores to each other in the "Datastore Mappings" tab within vSphere Replication.
- Create a single datastore and then create folders on that datastore for the individual hard disks the VM is using. e.g. For server FS1 that has 3 VMDKs, each on a different datastore you would create folders fs1_hd1, fs1_hd2, and fs1_hd3 on a single datastore. Then you can simply select those folders when setting up vSphere Replication for that VM.