Investigate flaky test-tls-fast-writing

[Trott]

Example failure: vs2015,win10

[Trott]

Stress test: https://ci.nodejs.org/job/node-stress-single-test/396/nodes=win10/console https://ci.nodejs.org/job/node-stress-single-test/397/nodes=win10/console

[Trott]

The original commit that landed this test seems to suggest that its purpose is to test that data is read again after drain fires.

The failure above seems to suggest that drain never fired (even though conn.write() had to return false at some point or the while loop would have locked up the thread and the test would have timed out).

This is the first time I've ever seen this test fail in CI (that I can recall, anyway).

And the stress test didn't result in any failures.

I'm inclined to close this. It can be re-opened if the test ever fails again, of course (or maybe if someone on @nodejs/platform-windows has any particular insights or interest in the issue).

[santigimeno]

@Trott I saw a TIMEOUT failure with this test in OS X. I can't seem to reproduce it though :(

[Trott]

@santigimeno reporting having seen this too, so reopening.

[Trott]

PR to increase the timeout: #5466

[BridgeAR]

Reopening due to a failure here:

https://ci.nodejs.org/job/node-test-commit-linux/15935/nodes=debian8-x86/console

  duration_ms: 0.169
  severity: fail
  stack: |-
  ...

[apapirovski]

I think this was a stackless error which is tracked elsewhere so I'm closing.