On Windows 2016 and windows 2012 we getting suite timeouts in what seem to be the very first suite run in that build.

It appears to that the test takes a long time between finishing the test method, and being marked as "done" - possibly due to clean up tasks in @After?

• https://elasticsearch-ci.elastic.co/job/elastic+elasticsearch+7.3+multijob-windows-compatibility/os=windows-2016/1/
• https://elasticsearch-ci.elastic.co/job/elastic+elasticsearch+7.3+multijob-windows-compatibility/os=windows-2012-r2/1/

Looking at the Windows-2016 failure:

• We start :server:test at 4:43 (04:43:11 > Task :server:test)
• We fail a little over 20 minutes (1,200,000 ms) later.

05:05:42 org.elasticsearch.node.NodeTests > testCloseOnLeakedStoreReference FAILED
05:05:42     java.lang.Exception: Test abandoned because suite timeout was reached
...
    java.lang.Exception: Suite timeout exceeded (>= 1200000 msec).

• The node being tested was stopped early in the 20 mminutes (7:45 in the node's TZ)

1> [2019-07-12T07:45:46,034][INFO ][o.e.n.Node               ] [testCloseOnLeakedStoreReference] closed 

• but the "after test" log doesn't come until we're killing off the suite (8:05)

1> [2019-07-12T08:05:47,854][INFO ][o.e.n.NodeTests          ] [testCloseOnLeakedStoreReference] after test

It seems that something on this new Ephemeral Windows CI worker is causing a problem moving this test method from "testing complete" to "test done"