New timeout impl leaves dangling threads

[headius]

This is discussed in detail in ruby/timeout#21. I will be disabling the testTimeoutCleanup test in our ScriptingContainerTest temporarily until we can get a fix merged and released.

[headius]

The test is still affected despite #7351 due to differences in how aggressively we shut down threads in an embedded scenario. Previously, we registered a hook that would shut down known timeout threads, as part of embedded teardown.

[justinstoller]

Where is this at?

Before I jump into attempting a local workaround, Is there a patch we take up or a preferred work around similar to puppetlabs/puppetserver@c376084 that we could employ?

[headius]

Newer JRuby versions have the fix from #7351, which makes timeout shut down properly under normal circumstances. Embedded JRuby use cases should get the same effect by tearing down the org.jruby.Ruby runtime. This does not fix the issue of the timeout library not providing a way to actively shut down without terminating the entire runtime, but it should prevent most typical users from seeing any issues.