[chef] RE: Re: tomcat6 start failing; not much info

["Kadel-Garcia, Nico" < >]

I'm afraid that the built-in tomcat6 init script isn't very good. It uses several different techniques in different parts of the script to detect if there is a running tomcat6, it has the root user create a /var/run/tomcat6.pid file owned by the tomcat6 user but leaves it empty, because only the tomcat6  daemon knows that its pid is, and if the "stop" command sent to the tomcat6 user as an "su" command does not succeed it does not go back and make sure that the daemon is dead. So the several tomcat6 "start" and "stop" steps in a row as a configuration is modified by chef can interfere with each other, because by the time the tomcat6 init script has completed, it's not *really* up and running and has left in place the empty, root generated pid file and not populated it yet.

Hilarity then ensues.

This is one of the reasons I dislike the tomcat6 cookbook, and in general dislike and building software packages from source tarballs: the RPM or apt bundled tomcat packages are, generally, much better about this sort of thing.

                                Nico Kadel-Garcia

---

From: Michael Hayes
Sent: Monday, November 04, 2013 7:47 PM
To:
Subject: [chef] Re: tomcat6 start failing; not much info

Thanks I will look at that. Unfortunately, this is just calling the off-the-shelf tomcat opscode-cookbook.

On Nov 4, 2013, at 6:11 PM, Daniel DeLeo < " target="_blank"> > wrote:

There’s not a lot of helpful information in this output. Doesn’t look like anything is timing out. You should look into the contents of the script to see if you can glean any insight. A common issue with java services is ENV vars not being set, so you could look at your own environment when you’re logged in to see if there’s any differences.

-- 

Daniel DeLeo