High swap usage(20%) on every node on large cluster in DGM

Setup
1. Create a 13 node cluster with 1 bucket, 1024 vBuckets
2. Load 51M items on the cluster [256 bytes - 512 bytes]
3. Mutate existing items and create new items [200 -612 bytes] to create around 60M items.

Output
1. The swap usage on every node suddenly shot up to 20%
2. Memory stats per node
Memory allocated - 6GB
Memory used - 4.34 G ( currently)
Available free RAM -1.64G
Swap usage - 1.16G

Stats/Resources/Screenshots
1.Attached are the memory stats from each node. https://s3.amazonaws.com/bugdb/jira/bug_swap/0529_swap.tar
2.Attaching the screenshot from the cluster
3. The cluster can be accessed at http://10.3.2.8:8091/index.html#sec=overview [Note : I ve done a [ add 2 nodes+ rebalance] on this, which failed so the stats might look a little different from the ones described above]
4. Attached the system logs at https://s3.amazonaws.com/bugdb/jira/bug_swap/syst.log.tar
Please let me know if you need any other information.