Forums

Home / Forums

You need to log in to create posts and topics. Login · Register

Solved: Migrating/Upgrading Cluster


alienn
37 Posts
May 20, 2019, 8:20 am

Hi,

I need to replace the hba/raid controller in my cluster nodes (3 node cluster) and I'd like to upgrade to Petasan 2.3 (now 2.2).
What would be recommended way of doing this in an active cluster?

I thought about this:

  1. upgrade cluster to 2.3
  2. shutdown first node and replace hba with raid controller (disk will be presented as jbod)
  3. run installer and choose "replace node"
  4. wait until everything settles and cluster is again in active/ok status
  5. repeat 2-4 with node 2 and 3

As the OSD names will surely change will I have to do anything about the then stale osds?
Has anyone done this before?

 

Cheers,
Nicki

Last edited on June 2, 2019, 6:27 pm by alienn · #1

admin
2,930 Posts
May 20, 2019, 10:24 am

If your controller will not require to re-format the disks (such as to raid volumes) , you should be able to just replace the controller and boot,  the drive labels may change but the  OSD ids do not depend on them (they are stored on the OSD disk) so it will not affect your OSDs.  If you OS disk is on the controller, you may need to select the boot disk in bios...but maybe i am missing something.

In this case, i would separate the controller change from the 2.3 upgrade, and do the first independently.

"Replace Management Node" is  done in the deployment wizard after you do a fresh install from the installer. If upgrading from the installer rather than a fresh install, you do not re-deploy the node

Last edited on May 20, 2019, 10:25 am by admin · #2

alienn
37 Posts
June 2, 2019, 6:27 pm

So... Replacing the hba with an areca controller went quite smooth and the update to 2.3 went perfect.

Now I'm facing the problem that I know what my bottleneck is and I quite don't understand it...

I'll open up another thread on this.

#3