I've had an issue with my preferred master IAP crashing when running 'show log provision', and noticed all clients connected on that cluster would lose connectivity for about 10 minutes (until the preferred master IAP came back). This bug brought my attention to how the VC changes IAPs, and what downtime it incurrs.
My understanding of the general process for VC changeover is: When the master IAP crashes or reboots, another IAP will adopt the VC. When the preferred master is back online the adopted VC will reboot, transferring the VC back to the preferred master.
I thought this process was quick, and would have my clients maintain their connection during the VC transition. However it currently takes the whole cluster down and takes some time to get all the clients reconnected. Is this normal, or likely related to a bug?