Hi,
one of our HA clusters (2-Node, CSS, 3TB) consists of two servers each with a RAID6 storage subsystem as data partitions for the img-files. (OS and Starwind Service are installed on a dedicated RAID1). Because I don't expect a high increase of data the next time I want to change the RAID6 layout to a Raid10 layout on each of both nodes.
My plan was to shutdown one node, backup up the data folder to an external USB3 hard disk drive and copy it back after the rebuild with the new layout will be finished and vice versa. Because this procedure should only trigger a fast sync in Starwind I guess I MUST guarantee the integrity of the data folder and I will not rely on a simple size check. My first idea was to run a hash check before I copy the data and after the restore. While the backup & restore part will run at ~130/160 MB/s the read speed during the hash calculation does not exceed 60MB/s, regardless of the algorithm. That would lead to a total duration of up to 3 days per node - too long for me.
So what about just forcing a full sync after the new layout on one server accomplished which should last much more less? This leads me to 2 questions:
1. How can I guarantee a full sync?
2. What files on the data partition on the "dirty" node must exist?
Thanks
The Latest Gartner® Magic Quadrant™Hyperconverged Infrastructure Software