ARL Headswap: Part 3/4 - Node C and Retiring Node B

on
Performing a non-disruptive ARL (Aggregate Relocate) Headswap for NetApp Clustered Data ONTAP - Step-by-Step Walkthrough Series:

Part 1 - Introduction and Preparation
Part 2 - Relocating Aggregates to Node B and Retiring Node A
Part 3 - Replace Node A with C, Relocating Aggregates to Node C and Retiring Node B
Part 4 - Replace Node B with D, Relocating Aggregates to Node D and Finish

Caveat Lector: Unofficial information!

10) Replace NODE-A with NODE-C

10.1) Cable NODE-C
Moving connections from NODE-A to NODE-C

10.2) Boot NODE-C to the Boot Menu
Power on NODE-C (if not already) and press Ctrl-C to access the boot loader environment.
At the LOADER> prompt type:

boot_ontap prompt

Interrupt the boot sequence by pressing Ctrl-C to get to the Boot Menu

10.3) Enter Maintenance Mode
At the Boot Menu, select (5) Maintenance mode boot

10.4) Verify MPHA *>

storage show disk -p


10.5) Acquire the System ID of NODE-C *>

sysconfig
disk show -a


10.6) Assign NODE-A’s Disks to NODE-C
Note: Only root aggr and spare disks get assigned to NODE-C. Check with ‘disk show -a’.
(8.2.x) Reassign disks from NODE-A to NODE-C *>

disk reassign -s NODE-A_SYSID -d NODE-C_SYSID

(8.3.x) Reassign disks from NODE-A to NODE-C *>

disk reassign -s NODE-A_SYSID -d NODE-C_SYSID -p NODE-B_SYSID

Note: The -p option is only required in maintenance mode when shared disks are present.
Enter n to ‘Abort reassignment (y/n)?’
Enter y to ‘After the node becomes operation ... Do you want to continue (y/n)?’
Enter y to ‘Disk ownership will be updated ... Do you want to continue (y/n)?’

10.7) Verify ha setting shows “ha” *>

ha-config show

Note: To modify use*>

ha-config modify controller ha
ha-config modify chassis ha


10.8) Destroy mailbox disks *>

mailbox destroy local
mailbox destroy partner


10.9) Verify personality of FC/UTA ports *>

ucadmin show
ucadmin modify


10.10) Exit maintenance mode *>

halt

This returns us to the LOADER> prompt.

10.11) Check Date
On NODE-B check the date in Clustershell::>

date

On NODE-C check the date and time at LOADER>

show date
show time

And if necessary>

set date mm/dd/yyyy
set time hh:mm:ss


10.12) Boot>

boot_ontap prompt


10.13) UPDATE FLASH FROM BACKUP CONFIG
IMPORTANT - DO NOT LET THE NODE BOOT WITHOUT DOING OPTION 6 FIRST!
Interrupt the boot by pressing Ctrl-C.
At the boot menu, select option (6) Update flash from backup config.

This will replace all flash-based configurations with the last backup to disks. Are you sure you want to continue? y

Enter y

The boot proceeds normally and the system then asks you to confirm the system ID mismatch. Confirm the mismatch.
Example:

WARNING: System id mismatch. This usually occurs when replacing CF or NVRAM cards! Override system id (y|n) ? [n] y

The node might go through one round of reboot before booting normally.

10.14) Verify NODE-C
Log in to NODE-C and from Clustershell::>

cluster show
event log show -messagename scsiblade.*
storage aggregate show -owner-name NODE-C


11) Return Data LIFs to NODE-C

11.1) Broadcast-Domains, Failover-Groups, VLANs, IFGRPs
Handle as required.

11.2) Return data LIFs to NODE-C
Typical commands::>

net int show -curr-node NODE-C
net int modify -vserver VSERVER -lif LIF_NAME -home-port PORT -home-node NODE-C
net int revert -vserver VSERVER -lif LIF_NAME


11.3) Restore SAN LIFs
Typical commands::>

net int show -curr-node NODE-C
net int modify -vserver VSERVER -lif LIF_NAME -status-admin up


12) Handle NODE-C’s Other LIFs

12.1) Cluster LIFs (Node Local)
Re-home as required (one LIF at a time).

12.2) Node-Mgmt LIF (Node Local)
Re-home as required.

12.3) Intercluster LIF (Node Local)
Re-home as required.

13) Use ARL to Relocate Aggregates from NODE-B to NODE-C

13.1) Perform Aggregate Relocation
Note 1: “-override-destination-checks true” is required since NODE-B and NODE-C are not a HA-pair.
Note 2: “-override-vetoes true” may be required
Relocate all the data aggregates from NODE-B to NODE-C::>

set adv
aggr show -owner-name NODE-B -root false

aggr relocation start -node NODE-B -destination NODE-C -aggregate-list AGGR_NAME -ndo-controller-upgrade true -override-destination-checks true
## REPEAT UNTIL ALL DATA AGGREGATES HAVE BEEN RELOCATED ##

storage aggregate relocation show -node NODE-B

Proceed once all data aggregates owned by NODE-B have been successfully relocated.

13.2) Verify Aggregates are Online and Check for Offline Volumes::>

storage aggregate show -nodes NODE-C -root false
volume show -node NODE-C -state offline


13.3) Verify “Node owns partner aggregates as part of the non-disruptive head upgrade procedure”::>

storage failover show -node NODE-C


13.4) Verify that NODE-B does not own any non-root aggregates that are online::>

storage aggregate show -owner-name NODE-B -ha-policy sfo -state online


14) Record NODE-B info

14.1) Networking Data Collection::>

network interface show -curr-node NODE-B
network port show -node NODE-B
network fcp adapter show -node NODE-B
ifgrp show -node NODE-B


14.2) Record/Verify SYSID (used later)::>

system node show -instance -node NODE-B


15) Migrate Data LIFs off NODE-B

15.1) Migrate data LIFs off NODE-B
Typical commands::>

net int show -curr-node NODE-B
net int modify -vserver VSERVER -lif LIF_NAME -home-port PORT -home-node NODENAME
net int revert -vserver VSERVER -lif LIF_NAME


15.2) Down SAN LIFs
Typical commands::>

net int show -curr-node NODE-B
net int modify -vserver VSERVER -lif LIF_NAME -status-admin down


16) Handle NODE-B’s Other LIFs and Networking

16.1) Cluster LIFs (Node Local)
It is important that at least one cluster port maps from the old platform to the new platform. Cluster LIFs would be temporarily homed on this port.

16.2) Cluster-Mgmt LIF (Cluster Wide)
If Cluster-Mgmt LIF(s) are on NODE-B then rehome to another node.

16.3) Node-Mgmt LIF (Node Local)
Ideally rehome to a port in-common to the old and new platform.

16.4) Intercluster LIF (Node Local)
Ideally rehome to a port in-common to the old and new platform.

16.5) Broadcast-Domains, Failover-Groups, VLANs, IFGRPs
Handle as required.

17) Retire NODE-B

17.1) Verify cluster status::>

set -c off; set adv; cluster show


17.2) (2-Node Cluster) Make sure epsilon is on NODE-C (which is staying up)::>

cluster modify -node NODE-B -epsilon false
cluster modify -node NODE-C -epsilon true


17.3) Verify no Data LIFs on NODE-B::>

net int show -role data -curr-node NODE-B


17.4) Halt NODE-B::>

system node halt -node NODE-B


17.5) Decommission NODE-B
- Wait for NODE-B to display the LOADER> prompt.
- (Dual-Chassis Configuration Only!) Power off
- Uncable NODE-B
- (Shared Chassis Configuration) Remove NODE-B from the chassis
- (Dual-Chassis Configuration) Remove NODE-B from the cabinet

17.6) (As required) Remove cards from NODE-B and install into NODE-D

Comments

Post a Comment