| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Using identical timeout values appears to have eliminated the flakiness in the test. Fixes: nodejs#7643
|
Argh... I don't feel good about it, but if it helps... LGTM |
Sorry, something went wrong.
|
Please disregard my LGTM |
Sorry, something went wrong.
|
After a bit of chat on IRC, LGTM again 💃 👍 |
Sorry, something went wrong.
Sorry, something went wrong.
Using identical timeout values appears to have eliminated the flakiness in the test. Fixes: nodejs#7643 PR-URL: nodejs#7717 Reviewed-By: Fedor Indutny <fedor.indutny@gmail.com>
| Back | FazBrowse Home | New Git URL |
Checklist
Affected core subsystem(s)
test http
Description of change
Using identical timeout values appears to have eliminated the flakiness in the test.
Fixes: #7643
With this change to the test and removing the fix in #6286, everything fails (as it should): https://ci.nodejs.org/job/node-stress-single-test/804/ (Test run just once on each platform)
With this change to the test only, everything passes (as it should): https://ci.nodejs.org/job/node-stress-single-test/805/ (Test run 100 times on each platform)
And with this change, the test is not flaky on FreeBSD:
https://ci.nodejs.org/job/node-stress-single-test/806/nodes=freebsd10-64/console (9999 runs, 0 failures)
For comparison, the test is flaky on current master on FreeBSD:
https://ci.nodejs.org/job/node-stress-single-test/807/nodes=freebsd10-64/console
(also 9999 runs, 1 failure, so not the overwhelming results I was hoping for, but at least it wasn't 0 failures...)
/cc @indutny
The flakiness is not terribly reproducible in the stress tests, but if it stands to reason that this reduces any race condition window, then it's probably worth doing.