Create a Post
cancel
Showing results for 
Search instead for 
Did you mean: 
Devilmac
Explorer
Jump to solution

Upgrading 3 members cluster

Hi,

 

I need to upgrade a 3 member cluster from R77.30 to R80.10

Until now i have been using the CP_Conectivity_Upgrade.

Where you upgrade first the two passive members to R80.10 and later the last one.

 

This time i need to upgrade one by one the members like this:

member1 - R77.30 active

member2 -R77.30 standby

member3 - R77.30 standby

to

member1 - R77.30 down

member2 - R77.30 down

member3 - R80.10 active

to 

member1 - R77.30 down

member2 - R80.10 active

member3 - R80.10 standby

to

member1 - R80.10 active

member2 - R80.10 standby

member3 - R80.10 standby

 

 

Do i need to do a cpstop in member2 then follow the CP_Conectivity_Upgrade with member1 and 3. And then upgrade member1 and 2 one by one

 

Is this correct?

 

Thank you!

0 Kudos
1 Solution

Accepted Solutions
Tommy_Forrest
Advisor

I've upgraded a number of 80.10 3-node clusters using the CU method and it has worked just fine.

However, the document I used did not exactly line up with 3 nodes.

With node 1 active, I did a cpstop on node 3.  Then I upgraded node 2.  Then I changed OS version in the cluster object and pushed policy with the "deploy to individual nodes" option.  Then forced the fault as the CU document noted.

Then I went and upgraded the other 2 nodes, pushed policy and every thing worked.

View solution in original post

0 Kudos
4 Replies
PhoneBoy
Admin
Admin
The more pertinent question is why are you upgrading to R80.10 and not R80.30, which is the widely recommended release? Even R80.20 would be a better choice here:
0 Kudos
Tommy_Forrest
Advisor

"The more pertinent question is why are you upgrading to R80.10 and not R80.30, which is the widely recommended release? Even R80.20 would be a better choice here:"

There could be a million reasons why they can't go to a higher version.  Their MDS being on 80.10  comes immediately to mind.

And perhaps they're not ready to upgrade to 80.20 or 80.30.

Sorry for being a little salty with this reply.  I spent 17 hours this past Saturday and 5 hours on Sunday trying to get to 80.30 only to have the installer tank our primary MDS and having to revert a 200 gig snapshot in the slowest possible manner.

 

 

0 Kudos
Tommy_Forrest
Advisor

I've upgraded a number of 80.10 3-node clusters using the CU method and it has worked just fine.

However, the document I used did not exactly line up with 3 nodes.

With node 1 active, I did a cpstop on node 3.  Then I upgraded node 2.  Then I changed OS version in the cluster object and pushed policy with the "deploy to individual nodes" option.  Then forced the fault as the CU document noted.

Then I went and upgraded the other 2 nodes, pushed policy and every thing worked.

0 Kudos
Devilmac
Explorer

This is what i needed to confirm.

Thanks!

0 Kudos

Leaderboard

Epsum factorial non deposit quid pro quo hic escorol.

Upcoming Events

    CheckMates Events