Upgrade Couchbase fail from 2.0.1 to 2.1.1


#1

I’m having this upgrade issue and I’m not sure how to move forward with it. I have 2 couchbase servers, server 1 and server 2. both in couchbase community 2.0.1. I brought the server 2 down and do the couchbase upgrade to 2.1.1. Upgrade successful but then I can’t restart the couchbase eventhough I rebooted already. Also for some reason after upgrade the chkconfig for couchbase disappear. Found out that there’s a bug for that particular issue so I ended up having to reinstall the whole couchbase server for server 2. No biggy. Installation successful. However when I tried to rejoin the cluster, server 2 gave me this error

Attention - Got no response from save request.

I’m sensing its because server 1 is still in 2.0.1 and it wouldn’t let new server 2 with 2.1.1 connect to it. Or is there a different issue? Anyone has ever went through this upgrade process with the same issue and has a solution for it? Please help?

FYI. Both my servers are on Centos 6.4 64 Bit.


#2

Are you following the online upgrade process in: http://docs.couchbase.com/couchbase-manual-2.2/#upgrading-to-couchbase-server-22 - specifically standard online upgrade - http://docs.couchbase.com/couchbase-manual-2.2/#standard-online-upgrades ? This should work fine, assuming you already joined the clutter with node 2 before you started upgrading.


#3

Hello,

Have you followed the suggestion from drigby ?

Regards
Tug


#4

Thanks drigby.

I believe that’s what I did. I was looking at my documentation on the steps I took and it looks exactly like what the document says.

However, since one of my server that I upgraded is already very screw up, I opt to try to just restore the server that I upgraded back to the old setting on 2.0.1. (Complete restore) Then when I try to add that server back to the cluster, it said that adding the new server to the cluster will remove the bucket in it, which is not good.

Let me try to reinstall that server with a brand new 2.0.1 and join the cluster and see whether that will work.

Now the other question before I do that is, I have the old server information still shown in the server nodes as down with the option to failover. How do I completely remove that old server? Should I completely remove that old server or couchbase will remove it after a certain amount of time? Or should I do the failover? I’m very confused right now. Please advise?


#5

Thanks for the follow up Tug. Sorry for the late reply. Crazy past 1 month… I just posted a comment above.


#6

Ok. I did follow the direction in the doc and I got tons of this error in the log:

Rebalance exited with reason {unexpected_exit,
{‘EXIT’,<0.9978.0>,
{badmatch,
[{‘EXIT’,
{{badmatch,{error,ehostunreach}},
{gen_server,call,
[<0.9977.0>,had_backfill,30000]}}}]}}}

that scared me so I cancel the removal. Is this normal and I can actually continue? Please advise?


#7

Let me add a caveat. I only have 2 nodes and when I try to upgrade one of the node, that’s the error I got. Is it because I have only 2 nodes?