How do I cancel a stuck upgrade?

[#1]

We have a customer who's grid upgrade got "stuck" when the active node of the grid master tried to upgrade. Passive node got upgraded ok and then failover occured, but the active node (which is now passive) did not reboot properly and now appears to be corrupt. All other nodes in the grid have been upgraded apart from this one.

Customer is trying to sort out an RMA but in the meantime is trying to add new nodes to the grid, however they get error messages stating that new nodes can't be added until the upgrade has been completed. Grid upgrade screen shows that the grid master has not completed upgrading - is there a way (maybe via a secret CLI command) to tell NIOS to abort the upgrade so that the customer can add these new nodes to the grid?

I'll log a ticket with support if no-one knows here, I just need to find out the serial number.

[#2]

Hi Paul,

I know this is a very old threat, and you probably already got your answer from Support. But for google search purposes I thought it would be useful to comment on this one.

And correct me if I'm wrong, but I thought this should be done by entering the following command in the CLI:

set grid_upgrade forced_end

A good post about this by adhill: https://community.infoblox.com/t5/Trending-KB-Articles/344-How-to-stop-Scheduled-upgrade-forcefully-...

Kind regards,

Jaip

[#3]

This only works for a scheduled upgrade, meaning and upgrade that did not start yet and is only scheduled.

If the upgrade is already in progress... it seems you are out of luck

[#4]

This answer might be very late, but adding this here for posterity and better Google search results.

From NIOS 8.6.3 onwards we have a new CLI command  “set grid_upgrade stop_and_cancel” to stop and cancel the upgrade status if we are stuck or in any similar scenario.

if you are in a Prior NIOS version, contact support with SSH access to GM active. Support can trigger similar results in the previous version by running a series of commands with root privileges.