ONTAP Discussions

shutdowm two node metrocluster cDOT

TR-3838
8,950 Views

Hello,

 

short question. I have a customer and he has a power maintenance in his datacenter. He ask how he can gracefull shut down a two node metrocluster without switchover and switchback ( both mcc are in the same dc for migration )

 

Any idea?

1 ACCEPTED SOLUTION

akiendl
8,938 Views

you are halting all nodes in the cluster on the site you want to shutdown:

 

halt -node * 

 

Keep in mind on boot you need to manually restart your SVMs.

https://kb.netapp.com/support/s/article/metrocluster-storage-virtual-machines-stop-after-a-cluster-reboot-or-an-unplanned-power-outage?t=1486028373531

 

ak.

View solution in original post

7 REPLIES 7

akiendl
8,939 Views

you are halting all nodes in the cluster on the site you want to shutdown:

 

halt -node * 

 

Keep in mind on boot you need to manually restart your SVMs.

https://kb.netapp.com/support/s/article/metrocluster-storage-virtual-machines-stop-after-a-cluster-reboot-or-an-unplanned-power-outage?t=1486028373531

 

ak.

niels
8,937 Views

Hi,

 

that would be:

 

cluster> node halt -node local -inhibit-takeover true

 

Issue the command on both clusters.

 

Which version of ONTAP are you running? This is important to understand how the SVMs would behave once you reboot the nodes.

 

regards, Niels

 

-----

If this post resolved your issue, please help others by selecting ACCEPT AS SOLUTION or adding a KUDO or both.

TR-3838
8,858 Views
Hello Niels, thanks for response. The Ontap Version is 9.1

niels
8,817 Views

In that case the SVMs should be in the running state after reboot.

If not please consult the KB article Andy provided earlier.

 

regards, Niels

LORENZO_CONTI
8,529 Views

Hello,

 

I have a two-node bridge-attached stretch metrocluster, which is the right procedure to:

1) power off node (cluster_A)
2) power off atto bridge and disk shelves that are in the same rack of the node

3) power on all equipment

This can be useful to move shelves or shelves+bridge to another rack.

Is it possible to do it without service disruption, or is better to power down both nodes?
Can we trust in planned metrocluster switchover?

Thank you 

Lorenzo

niels
8,504 Views

Hi Lorenzo,

 

yes - you can trust planned switchover. That's what it was designed for.

Please follow the link in order to get to the MetroCluster documentation that has more details on below steps.

https://library.netapp.com/ecm/ecm_download_file/ECMLP2495113

 

The whole procedure can be performed non-disruptively.

Here what I would suggest:

 

- perform switchover on the node you want to keep running - the other node will halt and can safely be powered off

- offline the plexes of all aggregates that reside on shelves you plan to power off

- power off bridges

- power off shelves

 

- relocate all components

 

- power on shelves - wait 2 min

- power on bridges - wait 2 min

- online the plexes - they will resync automatically

- wait for resync to complete

- perform "metrocluster heal aggregates"

- perform "metrocluster heal root"

- power on node

- perform switchback

 

 

Kind regards, Niels

 

-----

If this post resolved your issue, please help others by selecting ACCEPT AS SOLUTION or adding a KUDO or both.

 

LORENZO_CONTI
8,495 Views

Thanks Niels, it's more clear now 🙂


Public