Our scenario: We have a FAS2240-4 which was configured earlier with only one controller and the other controller was passive. The existing controller has 12 disks now and already assigned. Initially we bought this with 2 controllers but using one. Now we planned to add 12 more disks and enable the HA-pair(cf). I found a document from Netapp and I am following as per it. I have attached the document with this thread.
As per the document these are the steps:
1. Preparing for the upgrade on page 2 - ------- Done
2. Preparing a controller module when using Storage Encryption on page 2 ------ Not Applicable in our environment
3. Preparing the netboot server on page 3 -------- Done
4. Installing and cabling the new controller module on page 6 ------- Done
5. Booting and installing Data ONTAP on the new controller module on page 8 ----- Not Done (my question is here)
6. Setting up Data ONTAP on the new controller module on page 9
7. Verifying and setting the HA state of a FAS22xx, 32xx, or 62xx controller module and chassis on page 10
8. Enabling the cf.mode option or adding cf licenses on page 10
9. Running setup on the existing controller module to add the partner IP address on page 11
10. Rebooting both controller modules and enabling the HA pair on page 12
11. Installing the firmware after adding a second controller module on page 12
12. Cloning the configuration from the existing controller module to the new controller module on page 12
In the page 9 the instruction is as follows.
If your version of
Data ONTAP is...
8.x operating in 7-
a. Select the option Install new software first from the boot menu. ------- Done
This boot menu option downloads and installs the new Data ONTAP image to the boot device.
b. During the software installation process, when you are prompted for the URL of the image.tgz file, enter
the path as follows:
http://ip-address-of-existing controller/image.tgz ------- Done
After the software installation is complete, you are prompted to reboot the system.
c. Reboot the system. ------- Done
The system reboots and displays the boot menu.
d. Select option 4 to initialize all disks. -------- Not Done
When I select this option will it erases my data in my existing disks( 12 disks assigned to controller1 which is active now). If yes I will die
Also I didn't see any steps asking me to add the disks in the maintenance mode which one of my friend suggested before initializing (software ownership).
Could someone who have experienced in adding controller please share your thoughts. I am afraid of selecting this option before getting a clear idea.
Also please reply me ASAP as I have down time only today.
Thanks in advance.
Yes, you're correct that "option 4" will erase all content on all disks that are owned by the controller on which you run the command.
My expertise is in Clustered ONTAP so I am not quite sure of the procedure for 7-Mode. However I suspect that you can skip Option 4 unless you are adding disks that were previously used by a different environment/controller and you want to erase those disks before they are added to the environment.
I am not entirely sure how skipping this step affects the setup of ONTAP in 7-Mode or a HA pair in 7 Mode, so you may want to wait until someone familiar with 7-Mode answers.
You can manually manipulate disk ownerships in Maintenence Mode (option 5) with the "disk" and "storage" commands, if necessary. Option 4 won't touch disks that are not owned by that controller.
Thanks for your reply. The 12 disks which I have added is a new disk and no data in it and I am going to add it to ControllerB . But I was worried about the other 12 disks which is already in my ControllerA active and serving data . So my question is when I select the option 4, it wont delete my disks associated(software ownership) with my ControllerA. Am I right ?
Also My plan is to add the unowned disk to ControllerB then I will go for the initialization. Is that ok?
Option 4 won't touch any disks that aren't already owned by the controller you are running on. However to initialise the controller you probably do need to have some disks owned by it. Initialising/scrubbing those disks before you use them is a pretty good idea but can take several hours as the disks will be zeroed out (overwritten).
On Clustered ONTAP each controller needs at least 3 disks for a root aggregate before it can start up; so when I am installing a CDOT cluster I use maint mode to make sure that each node has at least 3 disks (and preferably a spare) before booting into ONTAP for the first time. To make the scrub faster I do not usually assign all disks at this time; later on I can change assignments for the other disks and run the remaining disk scrubs and RAID operations in the background while the filer is operating. I imagine the requirements for setting up a 7-Mode HA pair would be similar but I am not entirely sure. A fairly "safe" choice would be to assign all the disks that the node will own using maint mode, then use option 4 to scrub them before booting into ONTAP for the first time. You will of course need to make some decisions about how those disks are used (aggregates, RAID groups, vol0 and the like) so I recommend you read the relevant documentation or get advice from someone who knows 7-Mode better than I do.
You can boot into maint mode check and set the ownerships. Here's some commands to do that:
disk show -a
# Shows System ID of the current controller, and ownership of all attached disks. If the node hostnames are known, then you should also see the disk ownerships by hostname, otherwise you will only see the System IDs. The disk ownership info is recorded on a per-disk basis. The disk names are in the format of e.g. 0a.10.01, where 0a is the SAS "out" port of the controller, 10 is the shelf ID, and 01 is the disk number on the shelf.
disk show unowned
# Shows any unowned disks visible to the controller.
disk remove_ownership <disk id>
# Removes ownership info from a disk, you probably don't want this
disk assign <disk id> -s <new owner SysID>
# Assigns an individual disk to a controller
disk assign -shelf <shelf ID> -s <new owner SysID>
# Assigns an entire shelf to a controller; the shelf names are in the format 0a.shelf10 (for shelf with shelf ID 10 on SAS port 0a).
storage show shelf
# Shows the shelf IDs
# These two commands can help you check shelf connectivity for SAS shelves. If you have full multipath HA, you should have two paths to each shelf. You can use the serial number to identify the shelf. Note that on 22xx series, "shelf 0" is the internal disk bays.
Option 4 won't touch any disks that aren't already owned by the controller you are running on.
That's what I thought as well. But apparently current Data ONTAP became more intelligent. Recently someone posted that when controller has no assigned disks, it will automatically assign required number of unowned disks.
That's said, I always prefer to assign disks manually to be absolutely sure of the result.
Each controller is responsible for own disc
you should be assign new disk on maintanence mode
run the dısk show -n command you will see local system id:for exmp. 11111111
if you run """ disk assign all -s 11111111 """"command all unowned disk assigned new controller
after assign unowned disk run disk show command again you will see assigned disk
after disk assign run the halt command
when controller boot up data ontap will see new disk and its create automaticly root volume
if its doesn't work you should be open maintanence mode and choose 4a
it's important if you choose 4 -this command just initialize assigned disk
but 4a initialize assigned disk and create flex root volume
Thanks for your support and assistance. I finished it successfully.
Hi Gotkas I didn't see 4a in my options it has only 4. So I have chosen 4 after adding the disks in the maintenance mode.
But I used 2 TB disks which takes nearly 8 to 12 hours to initialize. Now everything is up and running fine.
Once again thank you all.
Thanks and Regards,