Skip to content

ci(capture): failure feature-flag input for OTel-Demo corpus captures #172

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
jensholdgaard merged 2 commits into from
Jun 10, 2026
Merged

ci(capture): failure feature-flag input for OTel-Demo corpus captures #172

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9443aa4
ci(capture): failure feature-flag input for OTel-Demo corpus captures
jensholdgaard Jun 10, 2026
38e6d96
fix(ci): glob-safe flag parse + fail loudly on a whitespace-only fail…
jensholdgaard Jun 10, 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
66 changes: 64 additions & 2 deletions .github/workflows/capture-otel-demo-corpus.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,6 +45,10 @@ on:
description: Concurrent load-generator users (HTTP, browser traffic off — drives corpus volume)
default: '40'
required: false
failure_flags:
description: Space/comma-separated flagd failure flags to force on (e.g. adFailure paymentFailure) — empty leaves the demo's defaults
default: ''
required: false
release_tag:
description: If set, publish the corpus as assets on a GitHub release at this tag (e.g. corpus/otel-demo-v1)
default: ''
Expand Down Expand Up @@ -113,6 +117,61 @@ jobs:
echo '--- effective extras ---'
cat demo/src/otel-collector/otelcol-config-extras.yml

- name: Enable failure feature flags
if: inputs.failure_flags != ''
Comment thread
jensholdgaard marked this conversation as resolved.
env:
# Via `env:`, not `${{ … }}` splicing into the script —
# dispatcher-controlled input, same rule as DEMO_REF.
FAILURE_FLAGS: ${{ inputs.failure_flags }}
run: |
set -euxo pipefail
# The demo's default traffic emits no error logs; flagd
# failure flags (adFailure, paymentFailure, …) are the
# demo-native way to make services emit real errors under
# load. flagd bind-mounts demo/src/flagd and reads
# demo.flagd.json from it, so patching the file before
# bring-up is the whole injection — same idea as the
# collector-overlay step above.
cfg=demo/src/flagd/demo.flagd.json
# `read -ra` (not command substitution) so glob characters
# can't expand against the workspace before validation;
# commas normalise to spaces first. Count what is applied —
# a whitespace-only input must fail loudly rather than
# silently capturing five hours with no errors enabled.
IFS=' ' read -ra FLAGS <<< "${FAILURE_FLAGS//,/ }"
applied=0
for flag in "${FLAGS[@]}"; do
[ -n "$flag" ] || continue
if ! [[ "$flag" =~ ^[A-Za-z0-9_-]+$ ]]; then
echo "::error::invalid flag name '$flag' (expected [A-Za-z0-9_-]+)"
exit 1
fi
if ! jq -e --arg f "$flag" '.flags | has($f)' "$cfg" >/dev/null; then
echo "::error::flagd flag '$flag' not in $cfg at this demo ref; available: $(jq -r '.flags | keys | sort | join(", ")' "$cfg")"
exit 1
fi
# The failing variant is "on" for most flags; the
# fractional ones (e.g. paymentFailure at 2.2.0) have
# percentage variants instead, where "100%" is the
# always-fail one.
variant="$(jq -r --arg f "$flag" \
'.flags[$f].variants | if has("on") then "on" elif has("100%") then "100%" else empty end' "$cfg")"
if [ -z "$variant" ]; then
echo "::error::flag '$flag' has no on/100% variant; variants: $(jq -r --arg f "$flag" '.flags[$f].variants | keys | join(", ")' "$cfg")"
exit 1
fi
jq --arg f "$flag" --arg v "$variant" \
'.flags[$f].defaultVariant = $v' "$cfg" > "$cfg.tmp"
mv "$cfg.tmp" "$cfg"
applied=$((applied + 1))
done
if [ "$applied" -eq 0 ]; then
echo "::error::failure_flags was set but contained no flag names"
exit 1
fi
echo '--- effective defaultVariants ---'
jq '.flags | map_values(.defaultVariant)' "$cfg"

- name: Bring up the demo
working-directory: demo
run: |
Expand Down Expand Up @@ -232,6 +291,7 @@ jobs:
# template-injection vector (the same rule the clone +
# capture steps already follow).
DEMO_REF: ${{ inputs.demo_ref }}
FAILURE_FLAGS: ${{ inputs.failure_flags }}
run: |
set -euxo pipefail
cd "$GITHUB_WORKSPACE/captured"
Expand All @@ -241,6 +301,7 @@ jobs:
echo "# OTel Demo corpus capture"
echo
echo "- demo_ref: ${DEMO_REF}"
echo "- failure_flags: ${FAILURE_FLAGS:-none}"
echo "- records (lines): ${lines}"
echo "- bytes: ${bytes}"
echo
Expand Down Expand Up @@ -277,13 +338,14 @@ jobs:
RELEASE_TAG: ${{ inputs.release_tag }}
DEMO_REF: ${{ inputs.demo_ref }}
DURATION: ${{ inputs.duration_seconds }}
FAILURE_FLAGS: ${{ inputs.failure_flags }}
run: |
set -euxo pipefail
gzip -kf logs.jsonl # logs.jsonl.gz, keep the original
records="$(wc -l < logs.jsonl)"
bytes="$(stat -c%s logs.jsonl)"
notes="$(printf 'OTel Demo corpus capture\n\n- demo_ref: %s\n- locust_users: %s\n- duration_seconds: %s\n- source commit: %s\n- records (LogsData batches): %s\n- uncompressed bytes: %s\n' \
"$DEMO_REF" "$LOCUST_USERS" "$DURATION" "$GITHUB_SHA" "$records" "$bytes")"
notes="$(printf 'OTel Demo corpus capture\n\n- demo_ref: %s\n- locust_users: %s\n- duration_seconds: %s\n- failure_flags: %s\n- source commit: %s\n- records (LogsData batches): %s\n- uncompressed bytes: %s\n' \
"$DEMO_REF" "$LOCUST_USERS" "$DURATION" "${FAILURE_FLAGS:-none}" "$GITHUB_SHA" "$records" "$bytes")"
# Create the release if it's new, else clobber its
# assets — re-minting the same tag replaces the corpus
# in place rather than erroring. On re-mint the notes
Expand Down
Loading