test(generator): усилены проверки startup-history и Superset
- Зачем: - зелёный результат проверок должен означать фактический стык и точный контракт дашборда. - Что: - проверка startup-history читает manifest, требует live seam и непустой ODS. - добавлен быстрый runtime gate для daily-wave и live-продолжения. - Superset sync ограничен целевым dashboard и сверяет существенные params. - Проверка: - docker target tests; make generator-test; make generated-history-runtime-check.
This commit is contained in:
@@ -12,8 +12,9 @@ CLICKHOUSE_PASSWORD="${CLICKHOUSE_PASSWORD:-123456}"
|
||||
GEN_MODEL_T0="${GEN_MODEL_T0:-2026-01-01T00:00:00+00:00}"
|
||||
GEN_MODEL_T_END="${GEN_MODEL_T_END:-2026-01-01T06:00:00+00:00}"
|
||||
REQUIRE_SUPERSET="${REQUIRE_SUPERSET:-1}"
|
||||
CHECK_LIVE_SEAM="${CHECK_LIVE_SEAM:-auto}"
|
||||
CHECK_LIVE_SEAM="${CHECK_LIVE_SEAM:-1}"
|
||||
GEN_LIVE_CHECK_MINUTES="${GEN_LIVE_CHECK_MINUTES:-10}"
|
||||
WAIT_LIVE_ROWS_SECONDS="${WAIT_LIVE_ROWS_SECONDS:-10}"
|
||||
|
||||
fail() {
|
||||
echo "Ошибка: $*" >&2
|
||||
@@ -50,11 +51,39 @@ CH_MODEL_T_END="$(clickhouse_datetime_literal "${GEN_MODEL_T_END}")"
|
||||
|
||||
[[ "${GEN_LIVE_CHECK_MINUTES}" =~ ^[0-9]+$ ]] \
|
||||
|| fail "GEN_LIVE_CHECK_MINUTES должен быть целым числом минут"
|
||||
[[ "${WAIT_LIVE_ROWS_SECONDS}" =~ ^[0-9]+$ ]] \
|
||||
|| fail "WAIT_LIVE_ROWS_SECONDS должен быть целым числом секунд"
|
||||
case "${CHECK_LIVE_SEAM}" in
|
||||
0|1|auto) ;;
|
||||
*) fail "CHECK_LIVE_SEAM должен быть 0, 1 или auto" ;;
|
||||
esac
|
||||
|
||||
echo "=== Проверка startup-history manifest ==="
|
||||
|
||||
manifest_summary="$(${COMPOSE_BIN} run --rm --no-deps \
|
||||
generator python -m clickstream_generator.startup_history_artifact kafka-manifest-summary)"
|
||||
|
||||
IFS=$'\t' read -r manifest_events manifest_visits manifest_users manifest_min_ts manifest_max_ts manifest_model_t0 manifest_model_t_end manifest_launch_profile <<< "${manifest_summary}"
|
||||
|
||||
[[ "${manifest_events}" =~ ^[0-9]+$ ]] || fail "не удалось прочитать startup-history manifest из Kafka"
|
||||
[[ -n "${manifest_model_t0}" ]] || fail "startup-history manifest не содержит model_t0"
|
||||
[[ -n "${manifest_model_t_end}" ]] || fail "startup-history manifest не содержит model_t_end"
|
||||
|
||||
CH_MODEL_T0="$(clickhouse_datetime_literal "${manifest_model_t0}")"
|
||||
CH_MODEL_T_END="$(clickhouse_datetime_literal "${manifest_model_t_end}")"
|
||||
|
||||
expected_profile="${PROFILE:-${GEN_LAUNCH_PROFILE:-}}"
|
||||
if [[ -n "${expected_profile}" && -n "${manifest_launch_profile}" && "${expected_profile}" != "${manifest_launch_profile}" ]]; then
|
||||
fail "проверка запущена для профиля ${expected_profile}, но manifest от ${manifest_launch_profile}"
|
||||
fi
|
||||
|
||||
echo "manifest_events=${manifest_events}"
|
||||
echo "manifest_visits=${manifest_visits}"
|
||||
echo "manifest_users=${manifest_users}"
|
||||
echo "manifest_model_t0=${manifest_model_t0}"
|
||||
echo "manifest_model_t_end=${manifest_model_t_end}"
|
||||
echo "manifest_launch_profile=${manifest_launch_profile}"
|
||||
|
||||
echo "=== Проверка ClickHouse: данные генерации в DM ==="
|
||||
|
||||
stats_query="
|
||||
@@ -271,7 +300,7 @@ echo "monotonic_ok=${contains_monotonic_ok}"
|
||||
echo "confirmation_share=${contains_confirmation_share}"
|
||||
|
||||
should_check_live_seam="${CHECK_LIVE_SEAM}"
|
||||
if [[ "${should_check_live_seam}" == "auto" ]]; then
|
||||
if [[ "${should_check_live_seam}" != "0" ]]; then
|
||||
live_rows_query="
|
||||
WITH
|
||||
toDateTime64('${CH_MODEL_T_END}', 6) AS t_end,
|
||||
@@ -280,12 +309,25 @@ SELECT count()
|
||||
FROM dds.event
|
||||
WHERE event_ts >= t_end AND event_ts < t_live_end
|
||||
FORMAT TabSeparated"
|
||||
live_rows="$(ch_query "${live_rows_query}")"
|
||||
[[ "${live_rows}" =~ ^[0-9]+$ ]] || fail "не удалось прочитать live-строки DDS после GEN_MODEL_T_END"
|
||||
if (( live_rows > 0 )); then
|
||||
should_check_live_seam="1"
|
||||
else
|
||||
should_check_live_seam="0"
|
||||
live_rows=0
|
||||
deadline=$((SECONDS + WAIT_LIVE_ROWS_SECONDS))
|
||||
while true; do
|
||||
live_rows="$(ch_query "${live_rows_query}")"
|
||||
[[ "${live_rows}" =~ ^[0-9]+$ ]] || fail "не удалось прочитать live-строки DDS после GEN_MODEL_T_END"
|
||||
if (( live_rows > 0 || SECONDS >= deadline )); then
|
||||
break
|
||||
fi
|
||||
sleep 1
|
||||
done
|
||||
if [[ "${should_check_live_seam}" == "1" ]] && (( live_rows == 0 )); then
|
||||
fail "live-продолжение не записало строки в первые ${GEN_LIVE_CHECK_MINUTES} минут после manifest model_t_end"
|
||||
fi
|
||||
if [[ "${CHECK_LIVE_SEAM}" == "auto" ]]; then
|
||||
if (( live_rows > 0 )); then
|
||||
should_check_live_seam="1"
|
||||
else
|
||||
should_check_live_seam="0"
|
||||
fi
|
||||
fi
|
||||
fi
|
||||
|
||||
@@ -425,19 +467,35 @@ WITH
|
||||
)
|
||||
SELECT
|
||||
(SELECT count() FROM crossing) AS crossing_visits,
|
||||
(
|
||||
SELECT count()
|
||||
FROM ods.device_by_click AS d
|
||||
INNER JOIN crossing AS c ON c.click_id = d.click_id
|
||||
WHERE d.click_id IS NOT NULL
|
||||
) AS ods_device_rows,
|
||||
(
|
||||
SELECT count()
|
||||
FROM ods.geo_by_click AS g
|
||||
INNER JOIN crossing AS c ON c.click_id = g.click_id
|
||||
WHERE g.click_id IS NOT NULL
|
||||
) AS ods_geo_rows,
|
||||
(SELECT count() FROM device_conflicts) AS ods_device_conflicts,
|
||||
(SELECT count() FROM geo_conflicts) AS ods_geo_conflicts
|
||||
FORMAT TabSeparated"
|
||||
|
||||
ods_context="$(ch_query "${ods_context_query}")"
|
||||
IFS=$'\t' read -r ods_crossing_visits ods_device_conflicts ods_geo_conflicts <<< "${ods_context}"
|
||||
IFS=$'\t' read -r ods_crossing_visits ods_device_rows ods_geo_rows ods_device_conflicts ods_geo_conflicts <<< "${ods_context}"
|
||||
|
||||
[[ "${ods_crossing_visits}" =~ ^[0-9]+$ ]] || fail "не удалось прочитать ODS-проверку стыка"
|
||||
(( ods_crossing_visits == crossing_visits )) \
|
||||
|| fail "ODS и DDS нашли разное число переходящих визитов: ${ods_crossing_visits}/${crossing_visits}"
|
||||
(( ods_device_rows > 0 )) || fail "ODS device пустой на стыке"
|
||||
(( ods_geo_rows > 0 )) || fail "ODS geo пустой на стыке"
|
||||
(( ods_device_conflicts == 0 )) || fail "ODS device конфликтует на стыке: ${ods_device_conflicts}"
|
||||
(( ods_geo_conflicts == 0 )) || fail "ODS geo конфликтует на стыке: ${ods_geo_conflicts}"
|
||||
|
||||
echo "ods_device_rows=${ods_device_rows}"
|
||||
echo "ods_geo_rows=${ods_geo_rows}"
|
||||
echo "ods_device_conflicts=${ods_device_conflicts}"
|
||||
echo "ods_geo_conflicts=${ods_geo_conflicts}"
|
||||
else
|
||||
|
||||
@@ -105,6 +105,7 @@ ${COMPOSE_BIN} up -d --no-deps superset
|
||||
echo "Шаг 8: техническая проверка аналитического контура"
|
||||
GEN_MODEL_T0="${GEN_MODEL_T0}" \
|
||||
GEN_MODEL_T_END="${GEN_MODEL_T_END}" \
|
||||
CHECK_LIVE_SEAM=0 \
|
||||
COMPOSE_BIN="${COMPOSE_BIN}" \
|
||||
bash "${SCRIPT_DIR}/check_generated_analytics.sh"
|
||||
|
||||
|
||||
@@ -0,0 +1,152 @@
|
||||
#!/usr/bin/env bash
|
||||
#
|
||||
# Быстрый runtime gate для issue 17: проверяет, что daily-wave доходит до
|
||||
# manifest/check, а live-продолжение реально записывает стык backfill/live.
|
||||
|
||||
set -euo pipefail
|
||||
|
||||
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
|
||||
REPO_ROOT="$(cd "${SCRIPT_DIR}/.." && pwd)"
|
||||
|
||||
COMPOSE_BIN="${COMPOSE_BIN:-docker compose}"
|
||||
PROFILE="${PROFILE:-daily-wave}"
|
||||
GEN_HISTORY_DURATION="${GEN_HISTORY_DURATION:-1h}"
|
||||
WAIT_CLICKHOUSE_SECONDS="${WAIT_CLICKHOUSE_SECONDS:-60}"
|
||||
WAIT_STG_SECONDS="${WAIT_STG_SECONDS:-5}"
|
||||
LIVE_SECONDS="${LIVE_SECONDS:-25}"
|
||||
GEN_LIVE_CHECK_MINUTES="${GEN_LIVE_CHECK_MINUTES:-10}"
|
||||
WAIT_LIVE_ROWS_SECONDS="${WAIT_LIVE_ROWS_SECONDS:-2}"
|
||||
|
||||
cd "${REPO_ROOT}"
|
||||
|
||||
launch_env="$(
|
||||
PYTHONPATH="${REPO_ROOT}/generator/src" \
|
||||
uv run python -m clickstream_generator.launch \
|
||||
backfill \
|
||||
--profile "${PROFILE}" \
|
||||
--duration "${GEN_HISTORY_DURATION}"
|
||||
)"
|
||||
eval "${launch_env}"
|
||||
|
||||
wait_for_clickhouse() {
|
||||
local deadline
|
||||
deadline=$((SECONDS + WAIT_CLICKHOUSE_SECONDS))
|
||||
|
||||
until ${COMPOSE_BIN} exec -T clickhouse clickhouse-client \
|
||||
--user=default \
|
||||
--password=123456 \
|
||||
--query "SELECT 1" >/dev/null 2>&1; do
|
||||
if (( SECONDS >= deadline )); then
|
||||
echo "Ошибка: ClickHouse не ответил за ${WAIT_CLICKHOUSE_SECONDS} сек." >&2
|
||||
exit 1
|
||||
fi
|
||||
sleep 2
|
||||
done
|
||||
}
|
||||
|
||||
stg_total_rows() {
|
||||
${COMPOSE_BIN} exec -T clickhouse clickhouse-client \
|
||||
--user=default \
|
||||
--password=123456 \
|
||||
--query "
|
||||
SELECT
|
||||
(SELECT count() FROM stg.browser_raw)
|
||||
+ (SELECT count() FROM stg.location_raw)
|
||||
+ (SELECT count() FROM stg.device_raw)
|
||||
+ (SELECT count() FROM stg.geo_raw)
|
||||
FORMAT TabSeparated"
|
||||
}
|
||||
|
||||
cleanup_live_generator() {
|
||||
${COMPOSE_BIN} stop generator >/dev/null 2>&1 || true
|
||||
}
|
||||
trap cleanup_live_generator EXIT
|
||||
|
||||
echo "=== Быстрая runtime-проверка startup-history/live seam ==="
|
||||
echo "GEN_LAUNCH_PROFILE=${GEN_LAUNCH_PROFILE}"
|
||||
echo "GEN_HISTORY_DURATION=${GEN_HISTORY_DURATION}"
|
||||
echo "GEN_MODEL_T_END=${GEN_MODEL_T_END}"
|
||||
echo ""
|
||||
|
||||
echo "Шаг 0: очистка volumes ClickHouse/Kafka/state"
|
||||
${COMPOSE_BIN} --profile live-generator down -v --remove-orphans
|
||||
|
||||
echo "Шаг 1: запуск ClickHouse и Kafka"
|
||||
${COMPOSE_BIN} up -d clickhouse kafka
|
||||
wait_for_clickhouse
|
||||
|
||||
echo "Шаг 2: применение DDL"
|
||||
bash "${SCRIPT_DIR}/apply_clickhouse_ddl.sh"
|
||||
|
||||
echo "Шаг 3: сборка образа генератора"
|
||||
${COMPOSE_BIN} build generator
|
||||
|
||||
echo "Шаг 4: короткий backfill стартовой истории"
|
||||
${COMPOSE_BIN} run --rm --no-deps \
|
||||
-e GEN_RUN_MODE=backfill \
|
||||
-e GEN_STATE_RESET=true \
|
||||
-e GEN_LAUNCH_PROFILE="${GEN_LAUNCH_PROFILE}" \
|
||||
-e GEN_SEED="${GEN_SEED}" \
|
||||
-e GEN_MODEL_T0="${GEN_MODEL_T0}" \
|
||||
-e GEN_MODEL_T_END="${GEN_MODEL_T_END}" \
|
||||
-e GEN_MODEL_TIMEZONE="${GEN_MODEL_TIMEZONE}" \
|
||||
-e GEN_MODEL_TIME_SPEED="${GEN_MODEL_TIME_SPEED}" \
|
||||
-e GEN_TICK_SECONDS="${GEN_TICK_SECONDS}" \
|
||||
-e GEN_LAMBDA_BASE_PER_MIN="${GEN_LAMBDA_BASE_PER_MIN}" \
|
||||
-e GEN_JITTER_PCT="${GEN_JITTER_PCT}" \
|
||||
-e GEN_MIN_EVENTS_PER_TICK="${GEN_MIN_EVENTS_PER_TICK}" \
|
||||
-e GEN_MAX_EVENTS_PER_TICK="${GEN_MAX_EVENTS_PER_TICK}" \
|
||||
generator
|
||||
|
||||
echo "Шаг 5: первый batch STG -> ODS -> DDS -> DM"
|
||||
sleep "${WAIT_STG_SECONDS}"
|
||||
bash "${SCRIPT_DIR}/run_batch.sh"
|
||||
stg_rows_before_live="$(stg_total_rows)"
|
||||
|
||||
echo "Шаг 6: live-продолжение, ждём новые STG-строки до ${LIVE_SECONDS} сек."
|
||||
GEN_RUN_MODE=live \
|
||||
GEN_STATE_RESET=false \
|
||||
GEN_LAUNCH_PROFILE="${GEN_LAUNCH_PROFILE}" \
|
||||
GEN_SEED="${GEN_SEED}" \
|
||||
GEN_MODEL_T0="${GEN_MODEL_T0}" \
|
||||
GEN_MODEL_T_END="${GEN_MODEL_T_END}" \
|
||||
GEN_MODEL_TIMEZONE="${GEN_MODEL_TIMEZONE}" \
|
||||
GEN_MODEL_TIME_SPEED="${GEN_MODEL_TIME_SPEED}" \
|
||||
GEN_TICK_SECONDS="${GEN_TICK_SECONDS}" \
|
||||
GEN_LAMBDA_BASE_PER_MIN="${GEN_LAMBDA_BASE_PER_MIN}" \
|
||||
GEN_JITTER_PCT="${GEN_JITTER_PCT}" \
|
||||
GEN_MIN_EVENTS_PER_TICK="${GEN_MIN_EVENTS_PER_TICK}" \
|
||||
GEN_MAX_EVENTS_PER_TICK="${GEN_MAX_EVENTS_PER_TICK}" \
|
||||
${COMPOSE_BIN} up -d --build generator
|
||||
deadline=$((SECONDS + LIVE_SECONDS))
|
||||
while true; do
|
||||
stg_rows_after_live="$(stg_total_rows)"
|
||||
if (( stg_rows_after_live > stg_rows_before_live )); then
|
||||
echo "live_stg_rows_before=${stg_rows_before_live}"
|
||||
echo "live_stg_rows_after=${stg_rows_after_live}"
|
||||
break
|
||||
fi
|
||||
if (( SECONDS >= deadline )); then
|
||||
echo "Ошибка: live-продолжение не записало новые STG-строки за ${LIVE_SECONDS} сек." >&2
|
||||
exit 1
|
||||
fi
|
||||
sleep 1
|
||||
done
|
||||
cleanup_live_generator
|
||||
|
||||
echo "Шаг 7: второй batch после live"
|
||||
sleep "${WAIT_STG_SECONDS}"
|
||||
bash "${SCRIPT_DIR}/run_batch.sh"
|
||||
|
||||
echo "Шаг 8: проверка manifest/profile/live seam без Superset"
|
||||
PROFILE="${PROFILE}" \
|
||||
GEN_LAUNCH_PROFILE="${GEN_LAUNCH_PROFILE}" \
|
||||
GEN_LIVE_CHECK_MINUTES="${GEN_LIVE_CHECK_MINUTES}" \
|
||||
WAIT_LIVE_ROWS_SECONDS="${WAIT_LIVE_ROWS_SECONDS}" \
|
||||
CHECK_LIVE_SEAM=1 \
|
||||
REQUIRE_SUPERSET=0 \
|
||||
COMPOSE_BIN="${COMPOSE_BIN}" \
|
||||
bash "${SCRIPT_DIR}/check_generated_analytics.sh"
|
||||
|
||||
echo ""
|
||||
echo "Готово: runtime-проверка startup-history/live seam прошла."
|
||||
Reference in New Issue
Block a user