ONTAP upgrade

[SVHO]

Per user support recommendation for an upgrade, there will need to be a maintenance windows when CIFS is being used.  Eventhough we have two controllers (FAS2650), is this true?  

On a typical upgrade, how long does it take for two controllers?

Thanks,

TT

[schmitz_peter]

Hi TT,

> Per user support recommendation for an upgrade, there will need to be a maintenance windows when CIFS is being used.  Eventhough we have two controllers (FAS2650), is this true?  

As the nodes have to be taken over and shut down one after another CIFS sessions need to be terminated.

Have a look at this thread: 

> On a typical upgrade, how long does it take for two controllers?

Using NDU, probably not more than an hour.

Cheers

Peter

[SVHO]

I remember testing failover with support onsite and the node that rebooted was waiting to take back and some cifs sessions were opened.  Is this the command to force it back online?

storage failover modify -node nas-01 -enabled true

Also, from the download, it gives me four files.

ONTAP - 9.2P1 (Binary Page)

• • 92P1_q_image.tgz (820776635 bytes)  (I am assuming its this one?)
  • 92P1_q_netboot.tgz (820776635 bytes)
  • 92P1_q_nodar_image.tgz (820636858 bytes)
  • 92P1_q_nodar_netboot.tgz (820636858 bytes)

Thanks,
TT

[AlexDawson]

Yes, 92P1_q_image.tgz is the correct file.

The _netboot files are used for restoring ONTAP under rare situations, and 'noDARE' (for 'no Data At Rest Encryption') is only needed for some nations which may ban encryption or which encryption technologies may not be exported to.

It is necessary to modify failover back on if it has been turned off, but the correct command to actually perform the giveback in the situation where CIFS sessions are open is: 

storage failover giveback ‑ofnode nodename ‑override‑vetoes true

Keep in mind that this will terminate any CIFS sessions needed and may in rare cases cause file corruption.

[SVHO]

Thanks for the answer.  I did find the commands on this link later.  Our upgrade went sucessfully.

https://library.netapp.com/ecmdocs/ECMP1196905/html/GUID-489519BE-0841-49B7-92F2-2B3034A53BC2.html

Regarding "It is necessary to modify failover back on".  I was told by support when they were onsite setting up our device to have the option turned off just so that when a controller goes bad, you don't want it to attempt to go back online.  Anyways, how would I see what the configuration is set?

When we attempted using the cli to get info regarding the open cifs sessions, the command below doesn't work.  It doesn't know what the session ID is ( after issuing vserver cifs session show -continuously-available !Yes -instance). We later used a combination of MMC (Microsoft console).

vserver cifs session file show -session-id session_ID

Anyways, we decided to shutdown the CIFS services to end all sessions but restarted back on right after the upgrade.  Would it be fine if we had left the CIFS service at a stopped state and perform the upgrade?

Thanks,

TT

[AlexDawson]

Fine to turn off CIFS for the duration of the upgrade. 

To show failover status, run "storage failover show". It's an almost religious debate if it should be on or not. Our current recommendation is to turn it on. There is code to detect multiple failures and not giveback if it has been flipping back and forth.