Shard(s) become assigned when lots of documents are pushed in ES in a short interval of time

Hi All - I have a two node (each node runs one instance) ES cluster setup and the index has 5 shards with replication enabled (default setting).
During the course of my testing, I tried to push lots of documents in ES server and after a while I see that one shard becomes unassigned. I left it in this way for a few hours and the shard was still unassigned. The problem went away after I restarted the ES server. I am able to duplicate this problem and sometimes I see even two shards getting into unassigned status.

Any help will be highly appreciated. I have ensured that my two ES servers have the same version and the underlying node has same Java version too.

Amit