Why does elasticsearch always do a recovery on startup?

Question

elasticsearch

Hello.
There is elasticsearch 2.1 as single-node on one server.
There are only three indexes.
What do I do. I start elasticsearch. I'm waiting for the cluster health status to become yellow.
Next I do service elasticsearch stop. Then service elasticsearch start.
After that, elasticsearch takes about 7 minutes to start, while the log is written like this:

[2015-12-23 22:30:04,285][DEBUG][index.shard              ] [Angar the Screamer] [my_index][2] state: [CREATED]->[RECOVERING], reason [from store]
[2015-12-23 22:30:04,286][DEBUG][indices.cluster          ] [Angar the Screamer] [my_index][4] creating shard
[2015-12-23 22:30:04,286][DEBUG][index.shard              ] [Angar the Screamer] [my_index][2] starting recovery from shard_store ...
[2015-12-23 22:30:04,287][DEBUG][index                    ] [Angar the Screamer] [my_index] [my_index][4] loaded data path [/var/lib/elasticsearch/elasticsearch/nodes/0/indices/my_index/4], state path [/var/lib/elasticsearch/elasticsearch/nodes/0/indices/my_index/4]
[2015-12-23 22:30:04,287][DEBUG][index                    ] [Angar the Screamer] [my_index] [my_index][4] creating using an existing path [ShardPath{path=/var/lib/elasticsearch/elasticsearch/nodes/0/indices/my_index/4, indexUUID='oSeWJoPbTzKL8lpqDlgyMA', shard=[my_index][4]}]
[2015-12-23 22:30:04,287][DEBUG][index                    ] [Angar the Screamer] [my_index] creating shard_id [my_index][4]
[2015-12-23 22:30:04,299][DEBUG][index.store              ] [Angar the Screamer] [my_index][4] store stats are refreshed with refresh_interval [10s]
[2015-12-23 22:30:04,299][DEBUG][index.deletionpolicy     ] [Angar the Screamer] [my_index][4] Using [keep_only_last] deletion policy
[2015-12-23 22:30:04,300][DEBUG][index.shard              ] [Angar the Screamer] [my_index][4] using [tiered] merge mergePolicy with expunge_deletes_allowed[10.0], floor_segment[2mb], max_merge_at_once[10], max_merge_at_once_explicit[30], max_merged_segment[5gb], segments_per_
[2015-12-23 22:31:13,850][DEBUG][index.shard              ] [Angar the Screamer] [my_index][4] scheduling refresher every 1s
[2015-12-23 22:31:13,851][DEBUG][index.shard              ] [Angar the Screamer] [my_index][4] state: [RECOVERING]->[POST_RECOVERY], reason [post recovery from shard_store]
[2015-12-23 22:31:13,851][DEBUG][index.shard              ] [Angar the Screamer] [my_index][4] recovery completed from [shard_store], took [1.1m]

How can I speed up the launch of elasticsearch? After all, she does not fall, but stops as expected.