[windows] erlang in one node crashed after finish warmup 7 nodes cluster

Environment:
9 nodes windows 2008 R2 64-bit
10.3.4.127
10.3.4.132
10.3.4.133
10.3.4.134
10.3.4.135
10.3.4.136
10.3.4.137
10.3.4.138
10.3.4.139

Create cluster 8 nodes with build 2.5.0-1059
Create 2 buckets: default 1 repplica and sasl-2 with 2 replicas
Load 20+ million items to each buckets to make dgm to 90% each buckets
Add a node 139 => passed
Remove node 137 => passed
Do failover tests. System test client unreachable due to xen management hang.
Reboot the whole host with windows vms in this host.
After restart all windows vms, all nodes in cluster up and items were back.
Observer node 139 had cpu closes to 99%
Then node 139 crashed.
Check erl processes in nodes 139 but UI is not up
See erl_crash.dump is generating in bin dir of node 139.
Live node is available now 10.3.4.139