Downtime for changing active controller?

[joebrug52]

What is the expected downtime for doing a "Make Active" on a standby controller? Is it even noticeable to users working off VM's running on the nimble? I'm changing some networking around and would like to swap to the other controller as part of the move. AF3000. Also, I've done firmware upgrades midday with no real noticeable issues. Any reason to expect differently? I do note the prompt that warns "Dont do this unless Nimble Support tells you to" which causes me pause Thanks

[buzzsubash]

Controller failover is transparent to the host and no downtime is needed.

We have to make sure that the hosts connected to the array has HPE storage/nimble connection manager installed that takes care of multipathing settings and adapater timeout values.

[joebrug52]

Okay, it probably took 5-10 minutes for it to sort itself out, for quite a while I got a spinning blue circle.

I never lost any pings to the mgmt IP or to one of the iscsi IP's, which was good..

but as far as the dashboard goes. Controller B showed "Solo" for quite some time with Controller A showing "Stale". I assume this is all expected but I was a little worried for a while

[buzzsubash · Accepted solution]

Glad to know.

yes, that is expected behaviour. When a failover is initiated, the services from active controller get handed over to the standby and the original active controller goes for a reboot.

This is when the the other controller shows as solo mode. Stale is when the controller is up, but waiting for services to start.

Consider the controller as physical server running linux and roughly it takes same time as a host to reboot after POST/BIOS checks followed by kernel load.