I fixed another class of memory leaks in Peregrine so we can now run longer jobs.
There are still 1-2 races on in some situations which can result in segfaults which are REALLY annoying for longer jobs.
I think I'm going to rewrite out use of ChannelBuffers so that if the underlying mmap is closed that the buffer will throw an IOException instead of a segfault.
This should help out vastly in the future to track down this class of bugs when closing resources.
Also, here's a video of my new kitten! :)