Issue Upgrading Couhbase to 6.5 using swap rebalance


I have tried to upgrade my couchbase version from 6.0.1 to 6.5.1
I have 11 nodes ( 4 data, 4 index, 2 query, 1 eventing) and 1 data node has around 1TB data
I try using swap rebalance, but found it takes a very long time to transfer the data & indexes.
Is there any work around for this?

Hi @Han_Chris1,

There are some other options for upgrading:

A quote from the docs: “If the speed of an upgrade is a primary concern for your cluster, see Graceful Failover or Performing the Offline Upgrade.”

Just to add, the preferred option for upgrade of index service node is (assuming each index has a replica):

  1. Failover one index service node at a time
  2. Upgrade the software
  3. Recover the node using rebalance

This allows for index service node to resume indexing from the point of failover and is usually a very fast operation while still being completely online.

1 Like

Hi @matthew.groves & @deepkaran.salooja,

Thanks for your response.
My current condition is I’m using docker
Actually I’m still confused for the offline upgrade step, there’s no detail step by to upgrade, it redirect to install single node.
Last time I’ve tried to stop the node, change docker image directly inside docker-compose. (I assume this is the step for offline upgrade using docker), then during bring up data service, it can not be started, and almost all of my data is corrupt.
That’s why I choose swap rebalance, but it takes longer time.


Now I’m a little confused, because there’s a 9-step process for Offline Upgrade at the above link ( Is there a particular step that you’re having trouble with?

Hi @matthew.groves ,

For step number 8, didn’t have a detail step for upgrade, only redirect to

Okay, I will rephrase my question,
How to do offline upgrade using docker?



I don’t think we’re looking at the same page, because the step 8 I’m looking at links to

But nevertheless, I think the steps will be similar, as long as you’ve mapped Couchbase node-specific data to a local folder/volume of /opt/couchbase/var/ (as discussed here: I’ve not personally tried this, but after a backup, could you stop the old docker image, and start a new docker image using the same volume?

Hi @matthew.groves,

we’re lookong at the same page, in that link, move forward to step number 6, it will redirect to install instead of upgrade.

For docker part, I’ve tried that step previously. and the result is my data node can not be started, it’s like corrupt.
So I rename my var folder, then it can be started but my old data can not be recovered.