06-05-2017 05:38 AM
As far as I tested in the lab when I have done the upgrade revert, all the memebers in the grid have rebooted at the same time. The ping for every memeber was time out at once.
Is there any way I can set the delay between the members while rebooting so I can reduce the service outage time?
In the meantime while the grid master candidate promotion, we can schedule grid members to restart sequentially so I'm sure there are some ways.
Solved! Go to Solution.
06-05-2017 08:27 AM
The Grid Revert is indended to recover from a catastrophic event or defect that is discovered shortly after an upgrade or for other maintenance purposes. The Revert causes the Grid to also revert to the former data set which could be even more disruptive after a period of time where various data changes have occurred since the upgrade.
A Revert is abrupt and intended to reboot all members to the previous NIOS version on the alternate partiion as quickly as possible. For full management and control, all members must be running the same version of NIOS. There is not a method to stager or stage this behavior, as there is with the GMC to GM promotion.
There is an option for a single member Revert if within the 24 hour window after an upgrade has occurred. Depending on the scenario, this may be appropropriate if you need to test or validate an issue is resolved by reverting.
Outside of lab work where you may want to flip back to the previous version of NIOS or test other behaviors, the Revert option should really only be considered if there is a serious service impacting issue, under the guidance of Support and with the understanding that the Grid Revert event will be disruptive to the entire Grid.
I hope that helps!