The admin guide doesn't really describe what a re-sync does and when to use it. I've tried it before when I've run into sync issues with mtree replication but it asks to put in the mtree name and errors out when you use the one at the replication target. It doesn't appear to me to have any additional functionality than creating a new replication pair.
Those links are for directory replication not mtree replication. In there was a link to a good KB explaining what resync is. https://support.emc.com/kb/180467
So resync compares files between source and destination before queuing and replicating segments. It could be used as a replacement to initializing. Resync is not able to be used for mtree replication even though it shows up as an option in system manager. Makes sense to me now. Resync wont work with mtree replication being that it uses point-in-time snapshot replication and there isn't really a initialization phase.
Is the replication resync command complaining that the destination mtree already exists?
This is because for mtree replication the destination mtree shouldn't exist. This is the reason why you cannot implement mtree replication on the /backup directory since this exists on all DDRs by default.
There are two ways to get past this:
1. Change the destination mtree by:
Rename: This will allow you to create the mtree replication context with the same name and global dedupe will be used to figure out what data needs to be sent to the destination DDR
FastCopy: FastCopy the mtree to another mtree name
The replication resync does work.
You will notice that whenever you break a replication context that already has a sync point a snapshot will be created to the effect of xxxxx-RESYNC-RESERVE-xxxxx. If this is not deleted on both source and target the resync command can be used.
It uses this snapshot as the baseline to resync all changes.