Reloading a VSF cluster of Aruba switches, as in the firmware upgrade procedure discussed, is not obviously running straight forward. Since I tried to google a couple of times and really found no straight comment on what actually happens, here the summary of my findings.
Long story short. A warm boot with
reload does not result in a predictable successfull manner. Never. Nowhere. That`s it!
The often cited
vsf sequenced-reboot is only supported on the 5400 platform. So all the kind remarks for users with 2930F, 3810 or similar VSF enabled platforms are simply not helpful. My guess is, that this is somehow supported through the hardware redundancy built in the 5400s and particular its management plane. Platforms lacking this kind of equipment simply can not deliver.
The only approach that reliable works on all the VSF enabled switches is
boot system flash primary or whatever flash bank you stored the image to boot. Its as simple as that. Do this and it works reliable and predictable.
All the good advice of timed reloads or boots is not necessary helpful regarding the successful reboot of a VSF stack, since it has no built in validity checks, but puts in a piece of luck to have the systems behave like intended. … Although scheduling any reboot is probably a good idea for operational reasons.
But do yourself a favor and reboot the system in the appropriate way!