Hi,
not sure if this is the right thread because it kind of crosses both SQL and Windows Failover Clustering so apologies if its out of place.
I have a 6 node SQL 2016 cluster which uses availability groups and a SAN for its storage, unfortunately the storage/networking has exhibited a few minor outages which has taken a LUN offline on a server which is hosting an instance, of course it only needs to be brief (or sometimes stuck in a reconnecting state) and the instance fails over to another node, bringing hundreds of databases offline in the process. we have around 25-30 LUNS (5 - 6 instances, some require mountpoints because we ran out of drive letters, each instance has data, logs, temp, backup, each its own LUN).
Ideally i'd like this to be CSV so we could have 1 CSV per instance instead of 5-6 LUNS per instance, plus it allows seamless transition of the storage if ISCSI was lost from a node.
My question is, is there a way we can migrate from LUN based storage assigned to a role, to a CSV based storage, either without any downtime or very, very little. bear in mind we have hundreds of databases per instance, collectively across all instances over 1,500.
as far as i understand it, we have to reinstall SQL on each node but pointing the instance to the CSV volume instead, but i'm guessing this would not be possible with existing instance names because the old INSTANCE-1 would point to.... drive F: and the new install of INSTANCE-1 would point to the CSV mountpoint, i'm sure this wouldnt be a possible configuration so i'm guessing the new instance name would have to be called something like INSTANCE-1A and we have to manually migrate the DBs from the old one to the new one?
I also assume that it must be a reinstall since the original install programs the configuration to look at a drive letter, we can't simply move our assigned storage to available storage and then to cluster shared volume storage can we?
i'm really hoping there is something out there which provides seamless migration but i can't find anything so far.
many thanks for your time
Steve