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

kubernetes-sigs / inference-perf / 43937092050
87%

Build:
DEFAULT BRANCH: main
Ran 16 Nov 2026 07:24PM UTC
Jobs 1
Files 142
Run time 1min
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

16 Nov 2026 07:13PM UTC coverage: 86.553% (+0.01%) from 86.541%
43937092050

push

github

web-flow
feat(config): default conversation replay concurrency to num_conversations (#1050)

Part of #540

Item 2 of the issue: an omitted `concurrency_level` on a conversation
replay stage now
defaults to `data.conversation_replay.num_conversations`. Items 3 and 4
are untouched.

**Old behavior.** `load.stages[].concurrency_level` was required on
every `concurrent`
stage, so the obvious shape — one active conversation per blueprint —
had to repeat
`num_conversations` in every stage; leaving it out was a `Field
required` error inside the
union of stage types. #924 delivered only item 1: it rejects
non-concurrent load types and
any `concurrency_level` above `num_conversations`.

**New behavior.** With `data.type: conversation_replay` and `load.type:
concurrent`,
`concurrency_level` may be omitted and resolves to
`data.conversation_replay.num_conversations`.
Everything else is unchanged: other data generators still have to name
their concurrency,
other load types are untouched, an explicit value is never replaced, and
a value above
`num_conversations` is still rejected with the same message. The count
is copied into the
stage as written and the stage field coerces it, so a quoted count still
sizes the stage
and a count `data` rejects (`4.7`, `"four"`) is not rounded down into a
stage size the data
field would refuse.

**Changes:**
- `_default_conversation_replay_concurrency` fills the field in on the
raw input, before the
stage models are built, and `Config` runs it as a `mode="before"`
validator. Scoped to
conversation replay on `concurrent`; every other config is returned
untouched.
- It rewrites raw input only. A validated `DataConfig`/`LoadConfig`, and
any dump of one,
already carries `concurrency_level` on every stage, so the enum and
model branches are
  gone rather than half-supported.
- `read_config` applies the same helper before its early
`ConcurrentLoadStage(**stage)`
construction, which would otherwise reject the omitted field first, and
validates th... (continued)

29 of 31 new or added lines in 1 file covered. (93.55%)

1 existing line in 1 file now uncovered.

13910 of 16071 relevant lines covered (86.55%)

0.87 hits per line

Uncovered Changes

Lines Coverage ∆ File
2
97.99
-0.33% inference_perf/config/config.py

Coverage Regressions

Lines Coverage ∆ File
1
79.17
-1.04% inference_perf/datagen/base.py
Jobs
ID Job ID Ran Files Coverage
1 43937092050.1 16 Nov 2026 07:24PM UTC 142
86.55
GitHub Action Run
Source Files on build 43937092050
  • Tree
  • List 142
  • Changed 2
  • Source Changed 1
  • Coverage Changed 2
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #43937092050
  • 5d2456a7 on github
  • Prev Build on main (#43934581025)
  • Next Build on main (#43955908636)
  • Delete
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