node does not remove completely in upgrade cluster from 1.8.0 to 1.8.1 in windows

Install cocubase server 1.8.0 on 3 nodes 24, 25 and 73
Create defautl bucket and load 100K items to this bucket
Do offline upgrade this cluster to couchbase server 1.8.1-927
Install couchbase server 1.8.1-927 on node 23 with custom install path c:/data-a and custom data path c:/data
Add node 23 to cluster
Create 2 more buckets, one is sasl and onther is dedicated port and load 100K items to these buckets
Load additional 100K items to default bucket.
Remove node 24 out of cluster. After rebalance done, node 24 was out the cluster but not reeset back to initial setup. Any change in node 24 will see in log of cluster.
At node 24, the rebalance button was clickable. Click rebalance on node 24 and see rebalance running. I stop rebalance and see data lost in default bucket on the cluster (lost 35 K items over 200 K items)