Redis status and statistics database connection broken would cause 99% cpu usage

What steps will reproduce the problem?
1. Start a redis-server in localhost, as turn-server's user-statsdb:
./redis-server &

2. Start turn-server with statsdb enabled:
./turnserver -p 10001 -L 10.10.10.15 -a -f -v --no-cli -r localhost -O 
port=6379 

3. Kill or restart the redis-server started in "step 1":
kill

4. run top command, watch the cpu usage of turn-server:
top

What is the expected output? What do you see instead?
Expected output:
Turn server runs normally, and cpu usage is low.

Real output: 
The cpu usage is over 99% per thread:

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND                                                                                                                         
30877 xxxxxxxx  20   0  696m  18m 2424 R 99.4  0.5   0:22.79 turnserver         


30876 xxxxxxxx  20   0  696m  18m 2424 R 99.1  0.5   0:22.80 turnserver         


30878 xxxxxxxx  20   0  696m  18m 2424 R 99.1  0.5   0:22.72 turnserver 


What version of the product are you using? On what operating system?
Version: 
Version Coturn-4.4.5.1 'Ardee West'

Operating system:
Linux tmpserver 3.2.0-23-generic #36-Ubuntu SMP Tue Apr 10 20:39:51 UTC 2012 
x86_64 x86_64 x86_64 GNU/Linux

Please provide any additional information below.

Original issue reported on code.google.com by huxinb...@gmail.com on 4 May 2015 at 1:43

Reproduce the failure with the listed turnserver and redis-server commands, then inspect the Redis statistics-database connection handling in the C code. Restarting or killing Redis should not make turnserver consume nearly 100% CPU per thread; verify the behavior with top after the connection is lost.