we snapmirror a volume (containing 1 LUN) to a DR system (3140 - 3140) break the relationship and present that snapmirrrored LUN to a DR ESXi host - when doing this we are asked by VMware about signatures, the 3 options are:
Keep Existing Signature - Persistently mount the LUN (for example, mount LUN across reboots)
In our lab i tried the first option (keep existing structure....) but found that the LUN was only mounting on 1 ESXi host and it didnt show up on the others even with a refresh - so i went with option 2 (Assign a new structure) when doing this the datastore name was changed slightly and it appeared on all ESXi hosts as planned
the second part of the question is what is the affect of doing this resignature from a NetApp / LUN point of view ? i.e. would the new signature have an effect on snapmirror resync ? meaning that even after a minor change on the underlying LUN that the signature change would cause a snapmirror to need to do a new snapshot full copy rather than delta changes ?
For option 1, you simply need to forcibly mount the datastore (detected as snapshot) on all ESX hosts, as per the quoted VMware KB article.
I wouldn't use option 2, unless this is an actual DR fail-over. If you do this, however, a new signature will be written to LUN, so it will get replicated back during resync (only block-level changes, not full copy) & then you will have to assign a new signature again at Primary