Clean shutdown of influxd server

[mcastilho]

There was not a clean shutdown path for influxd run command. The startProfiler() method was listening to signals but was forcing an os.Exit().

With this PR, I am implementing signal trapping and allowing opportunity for server to clean shutdown, sending Close() calls to the Server object appropriately.

Upon first signal, a timeout of 30 seconds is started, and if the clean shutdown hasn't been clompleted yet, a hard shutdown is initialized.

[otoolep]

Thanks @mcastilho -- have you signed the CLA?

[mcastilho]

Yes. I did @otoolep

[otoolep]

Great, thanks @mcastilho -- I tested your patch, works well.

[otoolep]

Looks good to me.

@dgnorton ?

[otoolep]

[run] 2015/07/15 10:31:00 signal received, initializing clean shutdown...
[run] 2015/07/15 10:31:00 waiting for clean shutdown...
[snapshot] 2015/07/15 10:31:00 snapshot listener closed
[tcp] 2015/07/15 10:31:00 cluster service accept error: network connection closed
[shard-precreation] 2015/07/15 10:31:00 precreation service terminating
[retention] 2015/07/15 10:31:00 retention policy enforcement terminating
[retention] 2015/07/15 10:31:00 retention policy enforcement terminating
[run] 2015/07/15 10:31:00 server shutdown completed

[corylanou]

+1

[dgnorton]

+1 Thanks @mcastilho!