5
88%
dev: 88%

Ran 15 Jul 2016 03:32PM UTC

Files 42

Run time 4s

Badge

Embed ▾

Committed 15 Jul 2016 03:24PM UTC coverage: 85.532%. Remained the same

Job # TEST_TIMEOUT_SCALE=20

Build Type

push

travis-ci

Committed by

web-flow

Commit Message

Fix flaky test TestResponseClosedBeforeRequest (#438)

On the writer side, the last flush of the "closeRequest" byte would fail
with "mex has been shutdown". The following sequence of events was
causing the issue:

 - Flush calls flushFragment, which sends a frame with the "closeRequest" byte
 - The remote side receives the byte, and closes the response
 - Sender receives the last response frame, and shuts down the mex
 - Original goroutine in Flush calls newFragment (since previous
   fragment) has been sent
 - newFragment fails since the mex has been shutdown

Since this is a valid sequence of events, we ignore the error from
Flush when writing out the closeRequest byte. This issue is unlikely
to happen in production, since the network calls to send and receive
frames should take longer than scheduling the goroutine running Flush.

Run Details

3819 of 4465 relevant lines covered (85.53%)

3688.07 hits per line

uber / tchannel-go / 2253 / 5
88%
dev: 88%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 2253.5 (TEST_TIMEOUT_SCALE=20)

uber / tchannel-go / 2253 / 5 88% dev: 88%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 2253.5 (TEST_TIMEOUT_SCALE=20)

uber / tchannel-go / 2253 / 5
88%
dev: 88%

README BADGES
x