• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

LFDT-Panurus / panurus / 31050564571 / 10
85%
main: 85%

Build:
DEFAULT BRANCH: main
Ran 05 Aug 2026 10:31PM UTC
Files 477
Run time 35s
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

05 Aug 2026 09:53PM UTC coverage: 29.541% (+0.03%) from 29.516%
31050564571.10

push

github

AkramBitar
test: stop three timing-dependent tests from flaking

Fixes #2121.

TestNextDelay_MaxDelayCap measured wall clock gaps between retries and asserted
each one roughly doubled. At a 10ms initial delay, scheduler jitter on a loaded
runner is a large fraction of the interval, so a ~34ms late wake up was enough
to fail it.

nextDelay is a pure function, so the policy can be asserted directly instead.
The new tests in retry_internal_test.go cover exact growth, the cap, the no
backoff case, the jitter band and the non-negative guard, and run in 0.086s
against 131s for the old one.

The non-negative guard needed two tests rather than one: the constructor clamps
jitterFactor to 1.0 and nextDelay applies (1 +/- jf/2), so the floor through the
public API is half the base delay and the finalDelay < 0 branch can never fire
(confirmed with 600k draws across jf=1.0, 2.0 and 3.0, all clamped, zero hits).
One test pins that the constructor clamps; the other builds the struct directly
with jitterFactor 3.0, where about a sixth of draws would go negative and must
come back as initialDelay instead. Checked that it fails when the factor is put
back to 1.0, so it does exercise the branch.

TestSufficientTokensBigDenominationsManyReplicas asserted that no replica ever
exhausts its retries, which is a scheduling outcome rather than an invariant.
Three replicas compete for two tokens there, yet it allowed 10 retries while the
single replica variant with no competition allows 20. Raised to 60 so contention
cannot decide the result.

TestGetRandomBytes_Success asserted the buffer is not all zeros for every
non-zero length, including the one byte case, where that just means the byte is 0
and so fails about 1 run in 256. Only assert it once the buffer is wide enough
for an all-zero draw to be implausible; randomness is already covered by
TestGetRandomBytes_Uniqueness.

Signed-off-by: atharrva01 <atharvaborade568@gmail.com>

12773 of 43238 relevant lines covered (29.54%)

42.73 hits per line

Source Files on job itest-fabtoken-fabric-t3 - 31050564571.10
  • Tree
  • List 477
  • Changed 4
  • Source Changed 0
  • Coverage Changed 4
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Build 31050564571
  • 8c6cf7e7 on github
  • Prev Job for on main (#31047221267.19)
  • Next Job for on main (#31071365638.21)
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc