NetApp head swap (FAS2020 / FAS3240) and upgrade (7.3.6 to 8.0.2)

Three years ago a client purchased a FAS2020, with 12 SAS disks in the head and a Fibre Channel shelf of ATA disks.
After a year (or so) they added a second FC shelf

The FAS2020 was a big success – but could not handle their new VDI project. There were performance issues surrounding running an office of 120 people (incl Exchange, SQL, Sharepoint – 13 servers total) and 60+ VDI desktops.

After crunching the numbers, we purchased a FAS3240 with one 24 disk SAS shelf and two 256GB PAM II cards.

Layout:
Old cluster
2020SAS (12 SAS disks in head)
2020SATA (24 SATA disks in two FC shelves)
2020 Network

New cluster
3240SAS (now- 21 SAS disks on one shelf. at the end – all 24 disk on one SAS shelf)
3240SATA (now- 3 SAS disks. at the end only the 24 SATA disks from 2020SATA)
Note the previous post on the network configs used:
3240 Network

Backups:
backup all the production data

On every filer:
rdfile /etc/rc
rdfile /etc/hosts
license
and copy all output.

run sysconfig -v
and note the system IDs for each head (you’ll want this for the disk reassign)

Start:
Run an ASUP on each cluster with a note:
options autosupport.doit Starting_swap

run cf disable on one head on each clusters

3240SATA
downgrade to 7.3.6
copy 736_setup_q.exe to \3240SATAc$etcsoftware
software update 736_setup_q.exe -r
reboot
Enter Maintenance Mode

Entering maintenance mode:
during a reboot:
hit control-c when prompted
type boot_ontap and hit enter
hit control_c when prompted
type 5 and hit enter to start maintenance mode

Remove the three temporary disks from the 3240SATA controller and add them to the 3240SAS controller
disk reassign -o 3240SATA -n 3240SAS
halt

on 3240SAS
note errors about foreign vol0 and aggr0 – both offlined and foreign aggr0 renamed to aggr0(1)
aggr destroy aggr0(1)

add the three disks to aggregate aggr0
aggr add aggr0 3

hot-removing shelves is not supported so:
on the 2020SAS
halt
then power off
2020SATA
halt
then power off
unplug the FC shelves

3240SATA
plug FC shelves into c0c

reboot
enter maint mode (see above)

Add all ATA disks to the filer
disk reassign -s old_sysid -d new_sysid

clear up the mailbox disks or 3240SATA will panic on boot
mailbox destroy local
mailbox destroy partner
halt
autoboot

manually set up filer on network
(it will boot with 2020SATA /etc/rc – I copied the /etc/rc from the test of the 3240SATA and rebooted)

copy proper* 8.0.2 software to \3240SATAc$etcsoftware
Note that the filer will boot with the CIFS config incl name from the 2020SATA.
(note that my filer shipped with 8.0.2P3 – but the download link for ONTAP 8.0.2 will give you the RTM, not P3. The p3 .zip didn’t contain p3 but also the RTM 8.0.2. the tgz p3 contained what I needed – so good luck with that!)

software update 802_setup_q.exe -r
reboot

test

add correct licenses license remove for invalid licenses, correct /etc/rc and /etc/hosts
double check options
If you have a PAMII you want to make sure flexcache.enable is on as that would be off coming from a 2020
on my config I had to set cifs.max_mpx 253 (note that regardless of the KB https://kb.netapp.com/support/index?page=content&id=3011249 saying the new default is 255 – ONTAP 8.x actually sets it to 253) as it was set to 64 on the 2020)

3240SAS
connect FC to c0c

disk show
cf enable

test incl failover/back
Note that CIFS will maintain the setup from the old filer – if you want to continue to use the old name and IP – don’t forget to add an alias so the new head also uses the old IP.

Run a “finished” ASUP when you’re done

options autosupport.doit Swap_Complete

eBay the old cluster