How do I add nodes (servers) with old data manually copied in?


#1

While swap rebalancing, 2 nodes went down from the cluster.
Initially, there were 3 nodes in the cluster. And I was trying to swap rebalancing 2 nodes (adding 2 nodes and removing 2 old nodes)

2 old nodes with bunch of data went down and never came back up.
I tried rebooting those 2 AWS EC2 instances, but the instances are NOT running back on.
So I had to failover the 2 nodes.

Now, I want to create 2 new instances and manually copy the bucket data from failed servers and add back to the cluster.
I copied the data under (…/var/lib/couchbase/data/mybucket).

But it seems like the cluster does not let me add these new nodes.

Rebalance exited with reason {buckets_cleanup_failed,
[‘ns_1@172.31.16.119’,‘ns_1@172.31.16.120’]}

Regardless of the error message, I don’t even want to cleanup the old data.
How could I properly do the job?

Any hint would be very appreciated.

Thank you.


Updated

Would cbtransfer solve the problem?


#2

You could use cbtransfer to take the data from the failed nodes and transfer it to the live cluster.

I’m not sure you could use it to take the data from the failed nodes and transfer it to the new nodes before they join the cluster, though. When you try add the new nodes to the cluster CB will want to wipe what’s on them and start the rebalance.

Do you mind me asking why you’re not happy to allow CB to populate the new nodes in a rebalance?


#3

Thank you @matthew

Well, I would love to CB to populate the new nodes in a rebalance for me.
However, 2 nodes went down while rebalancing and I can’t get them back up.
So I had to manually repopulate the date into the new 2 nodes.

Actually, I solved this issue with cbtransfer from file to cluster command.