Speed up sync after reboot of server and automatically start 'array'

[Benoire]

Hi

I've had some issues with power cuts and a failed UPS which forced my servers offline. I brought the first up and marked it as synced in order to start the VMs followed by the second server. After the bootup, the 2nd server took 8 hours to sync on 10gbe direct connection despite them both being turned off at roughly the same time. Do you checksum data to speed up the sync by comparing blocks on each server? Would make sense to establish what has changed since last boot rather than just sync all, certainly VMWare vSAN seems to only sync what it needs.

I am also assuming there is no automatic way for the vsan to start and mark itself as ready? Is the only way to create a script that basically marks one server (lets call it the primary) as ready and the starts the array? Clearly this is only a requirement after all servers have been rebooted as it shouldn't do this if one is online still otherwise you might end up with mismatched data.

Can Starwind vSAN automatically recover from reboots with the servers starting in a priority fashion, and then instead of syncing all again it only syncs the changes in files?

Thanks,

Chris

[Boris (staff)]

Chris,

StarWind VSAN is designed so that the partner services keep track of whether fast or full synchronization is required after any reboots of failures. Check this article for the reasons why full sync can be triggered.
As for the automatic recovery, in the current build this mechanism is already implemented, and usually the services need as little as 1ms to define the partner server loss. When the connection is restored between the nodes, they communicate with one another and check which one was switched off the latest. It gets marked as synchronized automatically. In some rare occasions, administrator interaction is still required to define which has to be marked as synchronized.