we are replacing old N5K switches with N9K.
Currently, we got a couple of FI connects to N5K vpc.Netapp storage connects to N5Ks
Plan is to move them to new N9Ks in a non-disruptive manner.
my migration plan is
- preconfigure N9Ks and connect them to existing N5Ks in a different vpc uplink(this will establish the connectivity path to netapp storage)
- during the migration window
- will put FI-B in maintenance mode
- shut down vpc 2 port channel on both N5Ks
- unplug the cables from N5K and connect them to N9K
- bring them up from N9Ks (no shut ports)
- bring them online from FI-B
- perform connectivity testing and if success
- migrate FI-A in a same manner.
Has any done FI migration in a similar way? were there any network outages?
my main concern is downtime as this has to be non-disruptive and no outage is expected.
Thanks,
Hi Ranga83,
I've migrated multiples times FIs to new upstream cluster switches. Depending on the upstream network policy which can help you out automatically while just disconnecting subordinate Upstream links which are:
- UCSM > LAN > Policies > root organization or sub-org > LAN Connectivity Policies: This one correctly configured will give you leverage to have One UCS disconnected from Upstream seamlessly by moving all data networks to currently active upstream FI - SW.
- UCSM > LAN > Policies > root organization or sub-org > Network Control Policies: by default, an existing policy will shutdown downstream ports if both FIs upstream ports are currently disconnected (which is not expected) but you set it up to 'Warning' instead of 'Link Down'. So, internal connectivity is not lost while FIs are eventually isolated.
If this doesn't help enough, you can use the regular Primary role switch and corresponding evacuation command, to move all active networking to one FI, so you can work with subordinate evacuated FI, move upstreams, and then switch back role and data by removing evacuation feature.
Traffic Evacuation section on Cisco UCS Manager Firmware Management Using the CLI, Release 4.1
AIV
