Status of some ledgers have remained IN_RECOVERY for long

**BUG REPORT**

#### What did you do?

We are using BookKeeper with [Pulsar](https://github.com/apache/incubator-pulsar).
Recently, the status of some ledgers have remained `IN_RECOVERY` for long in our cluster.

We tried to recover the ledgers by executing `readleger` command with `--force-recovery` option.
However, an exception occurred and the recovery failed.
https://gist.github.com/massakam/2136f3843259ec80af2d5f1312090854

During this time, the CPU usage of the Bookie server was very high.

#### What did you expect to see?

We expect that `IN_RECOVERY` ledgers change to `CLOSED` immediately and automatically.

#### What did you see instead?

The status of the ledgers does not change from `IN_RECOVERY`.

System configuration
BookKeeper version: 4.3.1.83-yahoo
Number of Bookies: 10
Ensemble size: 2
Write quorum size: 2
Ack quorum size: 2