Windows slave dying during build - unsatisfied link error

Windows slaved builds fail with the following error on the master:

11:56:20 FATAL: hudson.remoting.RequestAbortedException: java.io.IOException: Failed to abort
11:56:20 hudson.remoting.RequestAbortedException: hudson.remoting.RequestAbortedException: java.io.IOException: Failed to abort
11:56:20 at hudson.remoting.RequestAbortedException.wrapForRethrow(RequestAbortedException.java:41)
11:56:20 at hudson.remoting.RequestAbortedException.wrapForRethrow(RequestAbortedException.java:34)
11:56:20 at hudson.remoting.Request.call(Request.java:174)
11:56:20 at hudson.remoting.Channel.call(Channel.java:739)
11:56:20 at hudson.remoting.RemoteInvocationHandler.invoke(RemoteInvocationHandler.java:168)
11:56:20 at com.sun.proxy.$Proxy58.join(Unknown Source)
11:56:20 at hudson.Launcher$RemoteLauncher$ProcImpl.join(Launcher.java:956)
11:56:20 at hudson.tasks.CommandInterpreter.join(CommandInterpreter.java:137)
11:56:20 at hudson.tasks.CommandInterpreter.perform(CommandInterpreter.java:97)
11:56:20 at hudson.tasks.CommandInterpreter.perform(CommandInterpreter.java:66)
11:56:20 at hudson.tasks.BuildStepMonitor$1.perform(BuildStepMonitor.java:20)
11:56:20 at hudson.model.AbstractBuild$AbstractBuildExecution.perform(AbstractBuild.java:772)
11:56:20 at hudson.model.Build$BuildExecution.build(Build.java:199)
11:56:20 at hudson.model.Build$BuildExecution.doRun(Build.java:160)
11:56:20 at hudson.model.AbstractBuild$AbstractBuildExecution.run(AbstractBuild.java:535)
11:56:20 at hudson.model.Run.execute(Run.java:1740)
11:56:20 at hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:43)
11:56:20 at hudson.model.ResourceController.execute(ResourceController.java:88)
11:56:20 at hudson.model.Executor.run(Executor.java:234)
11:56:20 Caused by: hudson.remoting.RequestAbortedException: java.io.IOException: Failed to abort
11:56:20 at hudson.remoting.Request.abort(Request.java:299)
11:56:20 at hudson.remoting.Channel.terminate(Channel.java:802)
11:56:20 at hudson.remoting.Channel$2.terminate(Channel.java:483)
11:56:20 at hudson.remoting.AbstractByteArrayCommandTransport$1.terminate(AbstractByteArrayCommandTransport.java:72)
11:56:20 at org.jenkinsci.remoting.nio.NioChannelHub$NioTransport.abort(NioChannelHub.java:195)
11:56:20 at org.jenkinsci.remoting.nio.NioChannelHub.run(NioChannelHub.java:581)
11:56:20 at jenkins.util.ContextResettingExecutorService$1.run(ContextResettingExecutorService.java:28)
11:56:20 at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:439)
11:56:20 at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:303)
11:56:20 at java.util.concurrent.FutureTask.run(FutureTask.java:138)
11:56:20 at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:895)
11:56:20 at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:918)
11:56:20 at java.lang.Thread.run(Thread.java:695)
11:56:20 Caused by: java.io.IOException: Failed to abort
11:56:20 ... 9 more
11:56:20 Caused by: java.io.IOException: Connection reset by peer
11:56:20 at sun.nio.ch.FileDispatcher.read0(Native Method)
11:56:20 at sun.nio.ch.SocketDispatcher.read(SocketDispatcher.java:21)
11:56:20 at sun.nio.ch.IOUtil.readIntoNativeBuffer(IOUtil.java:198)
11:56:20 at sun.nio.ch.IOUtil.read(IOUtil.java:171)
11:56:20 at sun.nio.ch.SocketChannelImpl.read(SocketChannelImpl.java:245)
11:56:20 at org.jenkinsci.remoting.nio.FifoBuffer$Pointer.receive(FifoBuffer.java:136)
11:56:20 at org.jenkinsci.remoting.nio.FifoBuffer.receive(FifoBuffer.java:306)
11:56:20 at org.jenkinsci.remoting.nio.NioChannelHub.run(NioChannelHub.java:514)
11:56:20 ... 7 more

And the following warning on the slave agent:

Aug 19, 2014 11:51:09 AM hudson.util.ProcessTree get
WARNING: Failed to load winp. Reverting to the default
java.lang.UnsatisfiedLinkError: Native Library C:\Users\BuildWin\.jenkins\cache\ jars\12\winp.x64.FEF9CB80B43534DCA303AC36686258E8.dll already loaded in another classloader
at java.lang.ClassLoader.loadLibrary1(ClassLoader.java:1931)
at java.lang.ClassLoader.loadLibrary0(ClassLoader.java:1890)
at java.lang.ClassLoader.loadLibrary(ClassLoader.java:1851)
at java.lang.Runtime.load0(Runtime.java:795)
at java.lang.System.load(System.java:1062)
at org.jvnet.winp.Native.loadDll(Native.java:189)
at org.jvnet.winp.Native.load(Native.java:121)
at org.jvnet.winp.Native.<clinit>(Native.java:55)
at org.jvnet.winp.WinProcess.enableDebugPrivilege(WinProcess.java:200)
at hudson.util.ProcessTree$Windows.<clinit>(ProcessTree.java:469)
at hudson.util.ProcessTree.get(ProcessTree.java:328)
at hudson.Launcher$RemoteLauncher$KillTask.call(Launcher.java:930)
at hudson.Launcher$RemoteLauncher$KillTask.call(Launcher.java:921)
at hudson.remoting.UserRequest.perform(UserRequest.java:118)
at hudson.remoting.UserRequest.perform(UserRequest.java:48)
at hudson.remoting.Request$2.run(Request.java:328)
at hudson.remoting.InterceptingExecutorService$1.call(InterceptingExecut orService.java:72)
at java.util.concurrent.FutureTask.run(FutureTask.java:262)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor. java:1145)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor .java:615)
at hudson.remoting.Engine$1$1.run(Engine.java:63)
at java.lang.Thread.run(Thread.java:745)

Currently restarting the slave agent fixes the issue for a few builds but after about an hour the builds start failing again. It doesn't seem to be a memory issue as I've set the Xmx2048m parameter and the monitoring plugin seems fairly normal in it's reports.