Upstart Not Behaving as Configured

I have Logstash 1.4.2 freshly installed on Ubuntu Server 14.04 and using Oracle Java 1.7. I redirected log output to another directory instead of /var/log but failed to grant the proper permissions to the logstash user. This caused Logstash to immediately exit. I saw CPU usage peg because upstart was respawning Logstash as fast as it could.

/etc/init/logstash.conf has the stanza "respawn" with no parameters and no "respawn limit". According to http://upstart.ubuntu.com/cookbook/#respawn, in this case the default limit parameters "10 5" would apply, meaning that upstart should try to restart 10 times over a 5 second interval and then give up. This seems like a reasonable configuration. Instead, it's acting as if its respawn limit is "unlimited" as it never gives up retrying. This is not reasonable behavior as upstart is permanently stealing CPU cycles from other services and futilely wasting them.

This bug is a request to fix upstart's keep-alive behavior regarding Logstash so that it does not cripple the server when Logstash cannot run.