2
72%
master: 72%

Ran 07 Dec 2016 10:20AM UTC

Files 387

Run time 27s

Badge

Embed ▾

Committed 07 Dec 2016 10:09AM UTC coverage: 83.52% (+0.008%) from 83.512%

Job # 2770.2

Build Type

Pull #1453

travis-ci

Committed by

web-flow

Commit Message

pubsub: fix testStartStopReset

The test verifies that calling stop() actually stops the emulator.
The previous implementation stops the emulator by
- sending HTTP POST to /shutdown
- terminating the emulator process
The second step follows directly after the first.
It then queries the server it just shut down to make sure it does not
respond.

This behavior is incorrect for two reasons.
- The server might continue serving request after responding to
  /shutdown.
- The "emulator process" might not be the emulator at all.
  It might be a script that in-turn runs the emulator.
  On UNIX systems, killing a parent process does not kill any of its
  children; the actual emulator might run indefinitely after
  we kill the process we spawned.

This commit changes this behavior.
After POSTing to /shutdown, we simply wait for the spawned process
to terminate on its own.

Technically, there is still a race:
Another emulator could have been started on the same port.
When we send a request to verify that the old emulator has terminated,
we might get a response from the new one.
Since we execute tests sequentially though,
I believe this should not be a problem in practice.

Updates #1429.

Pull Request Pull Request #1453: pubsub: fix testStartStopReset

Run Details

16856 of 20182 relevant lines covered (83.52%)

122.04 hits per line

GoogleCloudPlatform / google-cloud-java / 2770 / 2
72%
master: 72%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 2770.2

GoogleCloudPlatform / google-cloud-java / 2770 / 2 72% master: 72%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 2770.2

GoogleCloudPlatform / google-cloud-java / 2770 / 2
72%
master: 72%

README BADGES
x