Skip to content

ci(recipes): use one load point for aggregated nightly tests #15571

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
sara4dev merged 4 commits into from
Oct 6, 2026
Merged

ci(recipes): use one load point for aggregated nightly tests #15571

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
c4b136b
ci(recipes): cap DeepSeek aggregated perf concurrency at 128
sara4dev Oct 2, 2026
93b208e
ci(recipes): use one DeepSeek aggregated perf concurrency
sara4dev Oct 2, 2026
188749b
ci(recipes): align Kimi aggregated nightly concurrency
sara4dev Oct 2, 2026
da2797f
ci(kimi): match nightly concurrency to trace sessions
sara4dev Oct 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion .github/ci/deepseek-v4-pro-agg/patch-perf.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,7 @@ spec:
- name: ENDPOINT
value: dsv4-pro-agg-frontend:8000
- name: CONCURRENCIES
value: "1 8 64 128 256 512 1024"
Comment thread
yunzhoul-nv marked this conversation as resolved.
value: "128"
Comment thread
sara4dev marked this conversation as resolved.
volumes:
- name: model-cache
persistentVolumeClaim:
Expand Down
4 changes: 2 additions & 2 deletions .github/ci/kimi-k25-agg/patch-perf.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,9 +21,9 @@ spec:
- name: perf
env:
- name: CONCURRENCIES
value: "1"
value: "20"
- name: BENCHMARK_DURATION
value: "60"
value: "120"
volumes:
- name: model-cache
persistentVolumeClaim:
Expand Down
8 changes: 5 additions & 3 deletions .github/workflows/shared-recipe-perf.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -257,15 +257,17 @@ jobs:
if: ${{ success() && !inputs.preflight_only }}
timeout-minutes: 5
run: |
concurrencies=$(yq -er '.spec.template.spec.containers[] | select(.name == strenv(PERF_CONTAINER)).env[] | select(.name == "CONCURRENCIES").value' "$ARTIFACTS/perf.yaml")
if [[ "$RECIPE" == kimi-k25-agg ]]; then
[[ "$concurrencies" =~ ^[1-9][0-9]*$ ]]
shopt -s nullglob
profiles=("$ARTIFACTS"/perf/*/*_trace_c1_*/profile_export_aiperf.json)
profiles=("$ARTIFACTS"/perf/*/*_trace_c"$concurrencies"_*/profile_export_aiperf.json)
(( ${#profiles[@]} == 1 ))
jq -e '.request_count.avg > 0 and (.error_request_count.avg // 0) == 0 and
# The CI trace has 20 conversations with 10 turns each.
jq -e '.request_count.avg == 200 and (.error_request_count.avg // 0) == 0 and
.output_sequence_length.avg > 0 and .was_cancelled != true' "${profiles[0]}"
exit 0
fi
concurrencies=$(yq -er '.spec.template.spec.containers[] | select(.name == strenv(PERF_CONTAINER)).env[] | select(.name == "CONCURRENCIES").value' "$ARTIFACTS/perf.yaml")
for concurrency in $concurrencies; do
jq -e --argjson requests "$((concurrency * 3))" '
.request_count.avg == $requests and (.error_request_count.avg // 0) == 0 and
Expand Down
Loading