What should we do before upgrade?
1) Download system files for current and desired version of Data ONTAP from MySupport pages
and upload it to /etc/software directory located on your filer.
2) Before upgrade unmount your exports with configuration files from servers and do the backup of /etc/hosts and /etc/rc
3) Choose the way of upgrade depending on your environment configuration:
|Type of SM||Upgrade method|
|Volume SnapMirror||Upgrade the storage systems containing the destination volumes, beginning with the furthest system in the topology (that is, the last system in a series of cascading destinations), and then upgrade the storage systems containing the source volumes.|
|Synchronous SnapMirroror semi-synchronous SnapMirror||Upgrade the storage systems containing the source volumes and the storage systems containing the destination volumes simultaneously.|
|Qtree SnapMirror||Upgrade the storage systems containing the source volumes before or after youupgrade the storage systems containing the destination volumes.|
Data ONTAP 7-Mode in High Availability pair – non-disruptive upgrade step by step
Just before you start with below plan ensure that everything on both filers are online (disks, volumes, aggregates, interfaces) and there are no maintenance jobs running like RAID scrubbing, reconstructions or disk wiping.
1) On both filers monitor the performance
filer1(2)> sysstat -c 30 -x 3
It’s good for a while to monitor the performance of our both nodes to ensure that CPU and Disk utilization is below 50% on each node. It’s because during takeover one node need to handle with both nodes. And low level of disk utilization is recommended to let ONTAP perform disk upgrade.
2) Trigger AutoSupport on both filers. Notification is send immediately and subject makes clear view on our actions
filer1(2)> options autosupport.doit "starting_NDU 8.X.XPX"
3) Verify status of HA pair from both filers
filer1(2)> cf status
4) Ensure that automatic giveback is off
filer1(2)> options cf.giveback.auto.enable off
5) Install the system files using the software command on both filers
filer1(2)> software update 8XXPX_q_image.tgz –r
6) Take care of snapmirror (optional step). Decide whether you want to wait until current transfers are finished. Below steps should be performed on both filers.
6.1) Look for any ongoing transfers, nice to have some management server to use grep
unix-mgmt> ssh filer01(02) snapmirror status |grep Transferring
If you don’t care whether current update finish job then go with
filer1(2)> snapmirror abort dest_vol
In other case use quiesce and wait for the whole update
filer1(2)> snapmirror quiesce dest_vol
If you know that more relationships have scheduled for next hour then do quiesce to them as well.
6.2) Turn off snapmirror
filer1(2)> snapmirror off
7) Terminate CIFS on the node to be taken over
filer2> cifs terminate
8) Time for takeover and giveback sequence on our HA pair. Be careful, statistics say the most failures happen in cluster configuration, because administrator become too far confident.
filer1> cf takeover
During takeover second filer goes down. Wait around 10 minutes to ensure that environment become stabilized, which means that clients multipathing and I/O request can be served after short pause caused because of takeover.
9) Ensure that filer2 is ready for giveback and give back the data service from filer1
filer1> cf giveback
Again wait some time before next step.
10) From the newly upgraded filer2, take over the data service from filer1
filer2> cf takeover
Once again wait some time before next step.
11) After the first node reboots and displays “waiting for giveback”, give back the data service from filer2
filer2> cf giveback
12) From filer1 and filer2 check the current version of ONTAP
13) Turn on automatic giveback, if that is your preferable setup
filer1(2)> options cf.giveback.auto.enable on
14) Check performance statistics
filer1(2)> sysstat -c 30 -x 3
And now probably you see high disk utilization. It means that your disk firmware is during update or some WAFL blocks reallocation has place. Anyway, it could take few hours before your system go back to fully operational status.
15) On both filers turn on snapmirror
filer1(2)> snapmirror on
and also quiesced SM has to be resumed
filer1(2)> snapmirror resume dest_vol
16) Trigger post-upgrade AutoSupport
filer1(2)> options autosupport.doit “finishing_NDU 8.X.XPX”
17) Verify the HA relationship is operational on both filers
filer1(2)> cf status
18) Confirm that all paths to disk are visible to the system by entering the following command on both filers
filer1(2)> storage show disk -p
19) Confirm that there are no failed disks by entering the following command on both filers
filer1(2)> vol status -f
20) Verify that all aggregates are online by entering the following command on both filers
filer1(2)> aggr status
21) Confirm that network interfaces are online by entering the following command on both filers
filer1(2)> ifconfig -a
What else is worth to know?
- Before you perform any implementation verify your plan with NetApp tool called Upgrade Advisor. In most cases it will help you determine the required steps. But be aware that you always have to think about your production environment to be not surprised in case of something goes wrong. That lead us to the next point, which is..
- Reverse plan – if you like your job, always ensure that you have backout plan. In terms of backout plan you have to turn off most of the features and make sure that the downgrade do not require any additional changes in system.
- Currently with firmware version of Data ONTAP, firmware for disks and disk shelves is bundled with system files. After filer reboot automatically firmware for disks and disk shelves is installed.