diff --git a/CHANGELOG.md b/CHANGELOG.md
index 523985d3..1415f9de 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -26,14 +26,27 @@
для smoke, отказов, persistence и failover.
- `scripts/restart.sh` перезапускает RabbitMQ и worker с сохранением volumes.
- `scripts/track1_experiment.py` и `make track1` — трек 1: матрица
- `serial/balanced/eager` × `small_many`/`huge_few`, throughput, queue lag,
- p95 time-to-ready, RSS и отдельный backpressure/recovery прогон;
+ `serial/balanced/eager` × `small_many`/`exactly_right`/`huge_few`,
+ throughput, queue lag, p95 time-to-ready, RSS, декомпозиция
+ decode/scale/JPEG/WebP и отдельный backpressure/recovery прогон;
self-contained HTML-дашборд с inline SVG.
+- `exactly_right` (1280×1280): optimized WebP кодируется без scaling, поэтому
+ encode share показывает нижнюю границу стоимости сжатия.
- Документация: README с запуском и архитектурой, отчёт с таблицей критериев,
CHANGELOG.
+### Removed
+
+- 34 устаревших файла `*.lab3`/`*.lab4` (Makefile, contracts, tests, scripts,
+ CI templates и fixtures) удалены из рабочего дерева.
+
### Changed
+- Параметры вариантов (`ThumbnailMax`, `CropSize`, `OptimizedMax`, качества
+ JPEG/WebP и WebP method) перенесены в `app/internal/config/config.go`.
+- Комментарии в Go/SQL/shell/Python больше не ссылаются на номера лабораторных;
+ поведение описывается по существу.
+
- `POST /chats/{id}/attachments` отвечает `202` после durable staging и COMMIT
задания; готовность определяется опросом метаданных, content до `ready` — 409.
- Загрузка потоковая (буфер 32 KiB), лимиты bytes/pixels проверяются до
diff --git a/README.md b/README.md
index bc3f00c6..402bbf64 100644
--- a/README.md
+++ b/README.md
@@ -150,35 +150,36 @@ python3 scripts/track1_experiment.py # трек 1 уров
`make test-persistence` / `make test-failover` с `scripts/restart.sh` и
`scripts/stop-one.sh`; см. `evidence/lab5-persistence*` и `evidence/lab5-failover.txt`.
-Трек 1 запускает матрицу профилей `serial/balanced/eager` на маленьких и
-больших изображениях, измеряет throughput, queue lag, p95 time-to-ready и RSS,
-затем отдельно показывает backpressure и восстановление. Результаты и графики —
+Трек 1 запускает матрицу `serial/balanced/eager` на трёх размерах:
+`small_many` (192×128), `exactly_right` (1280×1280, optimized WebP без scaling)
+и `huge_few` (2400×1800). Метрики: throughput, queue lag, p95 time-to-ready,
+RSS и разложение времени на decode / scale / JPEG / WebP. Результаты и графики —
`evidence/track1.{json,html,md}`, разбор — в [REPORT.md](REPORT.md).
```sh
-# Полный прогон: 30 small + 12 huge на каждый профиль, затем backpressure.
+# Полный прогон: 30 small + 18 exactly_right + 12 huge на профиль.
make track1 BASE_URL=http://localhost:8080
# То же без make.
python3 scripts/track1_experiment.py --base-url http://localhost:8080
# Быстрая проверка механики и/или свои лимиты worker-а.
-python3 scripts/track1_experiment.py --small-jobs 5 --huge-jobs 3 --skip-backpressure
+python3 scripts/track1_experiment.py --small-jobs 5 --exactly-jobs 3 --huge-jobs 3 --skip-backpressure
TRACK1_WORKER_CPU_LIMIT=2.0 TRACK1_WORKER_MEM_LIMIT=3G \
python3 scripts/track1_experiment.py --base-url http://localhost:8080
```
Эксперимент пересоздаёт worker с `WORKER_CONCURRENCY`/`WORKER_PREFETCH`,
-поднимает memory limit до 2 GiB (чтобы `eager` не падал по OOM на больших
-изображениях), а в `finally` возвращает конфигурацию из `.env`. Он управляет
-только локальным Compose-стендом и не предназначен для production.
+поднимает memory limit до 2 GiB (чтобы `eager` не падал по OOM), а в `finally`
+возвращает конфигурацию из `.env`. Он управляет только локальным Compose-стендом.
-Кратко: на 192×128 eager дал 13.86 jobs/s против 6.88 у serial (2×) при RSS
-28→33 MiB. На 2400×1800 все профили дали ~0.45 jobs/s (worker ограничен 1 CPU),
-но RSS вырос 205→1122 MiB, а queue lag p95 упал 23.3→17.5 s: concurrency меняет
-latency/ресурсы, а не throughput CPU-bound задач. При `MAX_PENDING_JOBS=3`
-четвёртый upload получил `503` + `Retry-After: 5`; после старта worker три
-принятых задания стали ready за 7.6 s.
+Кратко: на 192×128 eager дал 15.58 jobs/s против 6.84 у serial (2.3×) при RSS
+28→32 MiB. На 1280×1280 и 2400×1800 все профили дали ~0.44–0.50 jobs/s:
+worker ограничен 1 CPU, а encode share составил 82–83% и 61–64% — сжатие, а не
+только scaling, задаёт нижнюю границу. RSS вырос до 818–950 MiB, queue lag p95
+упал, но ready p95 остался ~26–36 s. При `MAX_PENDING_JOBS=3` API вернул
+`503` + `Retry-After: 5`; после старта worker принятые задания стали ready за
+7.65 s.
## Ограничения
diff --git a/REPORT.md b/REPORT.md
index ed8d1404..0921588c 100644
--- a/REPORT.md
+++ b/REPORT.md
@@ -115,7 +115,7 @@ backoff ±50% jitter до `JOB_MAX_ATTEMPTS`. После исчерпания
| Очистка staging после фиксации | сценарии `worker_stopped`, `crash_after_effect` | `staging_cleared: true`; reconciler удаляет только брошенное | `evidence/lab5-grade4.json`, `evidence/reconcile.txt` |
| Негативные случаи и лимиты | `python3 tests/custom/attachments.py` | 413 по bytes и по 42 MP, 415 corrupt/spoof, 404/400, нет 5xx | `evidence/lab5-negative-cases.txt` |
| Failover API | `make test-failover ACTION_SCRIPT=scripts/stop-one.sh` | 118 успешных записей, 1 переходная ошибка, восстановление 2.659 s; остановлена именно `api-2` | `evidence/lab5-failover.txt` |
-| Трек 1, уровень 5 | `python3 scripts/track1_experiment.py` | 3 профиля × 2 размера; eager/small 13.86 jobs/s против serial 6.88; huge_few ~0.45 jobs/s у всех, RSS 205→1122 MiB; cap=3 дал 503 + Retry-After 5 и восстановление за 7.596 s | `evidence/track1.json`, `evidence/track1.html` |
+| Трек 1, уровень 5 | `python3 scripts/track1_experiment.py` | 3 профиля × 3 размера; eager/small 15.58 jobs/s против serial 6.84; exactly_right/huge_few ~0.44–0.50 jobs/s, encode share 83% против 61–64%; RSS до 950 MiB; 503 + Retry-After 5, recovery 7.653 s | `evidence/track1.json`, `evidence/track1.html` |
## Эксперименты и отказы
@@ -143,42 +143,32 @@ HAProxy; отказ БД или брокера в этом сценарии не
### Гипотеза
-Маленькие изображения, которые не требуют масштабирования, дешевле на задание:
-выше throughput и ниже RSS. Большие изображения (2400×1800) требуют декодирования
-и downscale до 1280×960, поэтому очередь растёт, p95 time-to-ready увеличивается,
-а память сильнее зависит от concurrency. Дополнительно проверялся компромисс
-«latency против ресурсов»: больше concurrency/prefetch должно снижать queue lag,
-но повышать RSS.
+Стоимость задания = decode + scaling + encode. `exactly_right` (1280×1280) не
+масштабирует optimized WebP и показывает нижнюю границу стоимости сжатия;
+`huge_few` добавляет downscale. Ожидаем: на тяжёлых изображениях concurrency
+снижает queue lag, но не throughput при 1 CPU, а RSS растёт.
### Методика
-- Профили: `serial` 1/1, `balanced` 4/4, `eager` 8/32
- (concurrency/prefetch).
-- Сценарии: `small_many` — 30 PNG 192×128 (941 B, варианты не масштабируются);
- `huge_few` — 12 PNG 2400×1800 (53 135 B, 4.32 MP, optimized уменьшается до
- 1280×960).
-- Для каждой из 6 ячеек worker пересоздавался с профилем, поэтому RSS не
- переносился из предыдущего прогона. Изображения загружались параллельно
- (до 8 HTTP-клиентов). Все задания дожидались терминального состояния.
-- Метрики берутся из PostgreSQL по `attachment_jobs`: queue lag =
- `processing_started_at - created_at`, time-to-ready =
- `finished_at - created_at`, processing = `finished_at - processing_started_at`.
- Throughput = число ready / (последний `finished_at` − первый `created_at`).
-- RSS — `VmRSS` процесса worker из `/proc/1/status`, сэмпл примерно каждые
- 100 мс; baseline снимался до загрузки.
-- Worker ограничен **1 CPU**. На время эксперимента memory limit поднят до
- **2 GiB** (в `.env` по умолчанию 1 GiB), иначе `eager/huge_few` упирался бы в
- OOM, а не в измеряемый компромисс. Это зафиксировано в JSON.
-- Backpressure: API пересоздан с `MAX_PENDING_JOBS=3`, worker остановлен,
- upload-ы идут последовательно до первого `503`; затем worker запускается и
- проверяется восстановление принятых заданий и новый probe upload.
+- Профили: `serial` 1/1, `balanced` 4/4, `eager` 8/32 (concurrency/prefetch).
+- Сценарии: `small_many` 30×192×128; `exactly_right` 18×1280×1280 (optimized
+ WebP без scaling); `huge_few` 12×2400×1800 (optimized → 1280×960).
+- 9 ячеек; worker пересоздаётся на профиль, изображения загружаются параллельно,
+ все задания доводятся до терминального состояния.
+- Queue lag, time-to-ready и processing — из `attachment_jobs`; throughput =
+ ready / (последний `finished_at` − первый `created_at`).
+- Decode/scale/JPEG/WebP — из worker-лога; encode share =
+ `encode / (decode + scale + encode)`. Это wall-clock под данным профилем.
+- RSS — `VmRSS` из `/proc/1/status` примерно каждые 100 мс.
+- Worker 1 CPU, memory limit 2 GiB (в `.env` 1 GiB), чтобы `eager` не падал по
+ OOM; зафиксировано в JSON.
+- Backpressure: `MAX_PENDING_JOBS=3`, worker остановлен, upload-ы до `503`, затем
+ recovery. Cap выбран намеренно; важны `Retry-After` и отсутствие потерь.
Команда:
```sh
make track1 BASE_URL=http://localhost:8080
-# или
-python3 scripts/track1_experiment.py --base-url http://localhost:8080
```
Сырые данные: `evidence/track1.json`; графики: `evidence/track1.html`;
@@ -186,63 +176,46 @@ python3 scripts/track1_experiment.py --base-url http://localhost:8080
### Результаты
-| Профиль | Сценарий | Jobs/s | Queue lag p95, ms | Ready p95, ms | Processing p95, ms | RSS peak, MiB | Max parallel |
-| --- | --- | ---: | ---: | ---: | ---: | ---: | ---: |
-| serial | small_many | 6.875 | 3 194 | 3 284 | 113 | 28.2 | 1 |
-| serial | huge_few | 0.443 | 23 333 | 25 521 | 2 308 | 204.7 | 1 |
-| balanced | small_many | 11.590 | 1 624 | 1 736 | 209 | 30.6 | 4 |
-| balanced | huge_few | 0.454 | 17 716 | 25 785 | 25 374 | 728.7 | 4 |
-| eager | small_many | 13.858 | 898 | 1 158 | 284 | 32.7 | 8 |
-| eager | huge_few | 0.459 | 17 538 | 25 699 | 24 979 | 1 122.1 | 8 |
+| Профиль | Сценарий | Jobs/s | Queue lag p95, ms | Ready p95, ms | Processing p95, ms | Encode p95, ms | Encode share, % | RSS peak, MiB | Max parallel |
+| --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: |
+| serial | small_many | 6.837 | 3 256 | 3 338 | 170 | 80 | 83.8 | 28.2 | 1 |
+| serial | exactly_right | 0.475 | 33 591 | 35 714 | 2 180 | 1 722 | 82.0 | 174.3 | 1 |
+| serial | huge_few | 0.436 | 23 668 | 25 896 | 2 359 | 1 508 | 63.8 | 212.9 | 1 |
+| balanced | small_many | 13.694 | 1 147 | 1 274 | 396 | 90 | 83.8 | 28.9 | 4 |
+| balanced | exactly_right | 0.495 | 31 510 | 35 764 | 8 174 | 6 562 | 83.1 | 433.8 | 4 |
+| balanced | huge_few | 0.441 | 18 161 | 26 576 | 9 046 | 5 606 | 61.4 | 756.4 | 4 |
+| eager | small_many | 15.576 | 924 | 1 166 | 410 | 95 | 79.9 | 32.1 | 8 |
+| eager | exactly_right | 0.495 | 31 735 | 35 797 | 16 363 | 13 330 | 83.1 | 818.4 | 8 |
+| eager | huge_few | 0.443 | 17 648 | 26 641 | 17 678 | 11 124 | 61.4 | 949.6 | 8 |
-Backpressure: cap `3`; первые 3 upload-а получили `202`, четвёртый — `503` с
-`Retry-After: 5` и `error_code=unavailable`. После старта worker три принятых
-задания стали `ready` за **7.596 s** (включая запуск контейнера и healthcheck),
-после чего probe upload получил `202` и дошёл до `ready`.
+Backpressure: cap `3` задан намеренно; первый `503` пришёл с `Retry-After: 5`.
+После старта worker принятые задания стали `ready` за **7.653 s** (включая
+запуск контейнера и healthcheck), probe upload получил `202` и дошёл до `ready`.
### Выводы
-1. **Размер изображения важнее числа заданий.** При `balanced` маленькие файлы
- дают 11.59 jobs/s, большие — 0.454 jobs/s (в 25.5× меньше); p95 time-to-ready
- растёт с 1 736 до 25 785 ms (в 14.9×). Processing p50 для huge — 8 460 ms
- против 211 ms для small: доминирует декодирование/масштабирование, а не
- ожидание в очереди.
-2. **Eager хорошо работает на маленьких заданиях.** 13.86 jobs/s против 6.88 у
- serial (2.0×), queue lag p95 падает с 3 194 до 898 ms, RSS растёт всего с
- 28.2 до 32.7 MiB. Это ожидаемый выигрыш от перекрытия лёгких задач.
-3. **На huge_few concurrency не даёт throughput.** Все три профиля дают
- ~0.44–0.46 jobs/s: worker упирается в 1 CPU, а декодирование и WebP-кодирование
- уже насыщают его. Queue lag p95 снижается (23.3 → 17.5 s), но p95
- time-to-ready остаётся ~25.5 s: последние задания всё равно ждут своей доли
- CPU. При этом RSS растёт с 205 MiB (serial) до 729 MiB (balanced) и
- 1 122 MiB (eager). **Больше concurrency здесь — не бесплатное ускорение, а
- рост памяти при том же throughput.**
-4. **Backpressure ограничивает принятую работу.** При cap=3 API перестал
- принимать новые файлы до чтения тела и вернул retryable `503`; уже
- подтверждённые задания не потерялись и завершились после возврата worker.
- Это и есть требуемое «bounded acceptance + recovery».
-5. **Практический выбор.** Для лёгких изображений разумен `balanced`/`eager`;
- для тяжёлых при одном CPU лучше `balanced` (или `serial`, если важна память):
- throughput тот же, а RSS в 1.5–5.5× ниже, чем у eager. `prefetch` больше
- `concurrency` лишь держит сообщения у worker-а и не увеличивает параллельную
- обработку.
+1. **Compression, not scaling, is the floor.** `exactly_right` даёт 82–83%
+ encode share и ~0.48 jobs/s; `huge_few` — 61–64% и ~0.44 jobs/s. Убрать
+ downscale недостаточно, чтобы 1280×1280 стал дешёвым: доминирует WebP.
+ На `huge_few` scale p95 выше (balanced: 3.4 s против 1.5 s), но optimized
+ output меньше (1.23 MP против 1.64 MP), поэтому throughput почти совпадает.
+2. **Маленькие изображения масштабируются concurrency.** 6.84 → 13.69 →
+ 15.58 jobs/s, RSS 28 → 32 MiB.
+3. **Тяжёлые изображения CPU-bound.** ~0.44–0.50 jobs/s у всех профилей;
+ queue lag падает, ready p95 держится ~26–36 s, RSS растёт 174 → 818 MiB
+ (`exactly_right`) и 213 → 950 MiB (`huge_few`). Concurrency покупает latency
+ ценой памяти, а не throughput.
+4. **Backpressure.** `503` + `Retry-After: 5`; принятые задания восстановились
+ за 7.65 s.
+5. **Практический выбор.** `balanced` для тяжёлых, `eager` для лёгких.
-### Ограничения эксперимента
+### Ограничения
-- Один worker, один RabbitMQ, один PostgreSQL; это сравнение настроек, не HA и
- не горизонтальное масштабирование.
-- Worker ограничен 1 CPU и 2 GiB; на другом железе абсолютные числа и точка
- насыщения будут другими. Относительный вывод «CPU-bound huge не ускоряется
- concurrency» получен именно для этого лимита.
-- PNG-фикстуры синтетические и хорошо сжимаются; доминирует стоимость
- декодирования/масштабирования, а не сеть. Размеры в байтах и SHA-256 есть в
- JSON.
-- Один прогон на ячейку, без длительного прогрева и повторений; p95 по 12
- заданиям чувствителен к выбросу. Это учебная оценка, не SLA.
-- `MAX_PENDING_JOBS` проверяется приблизительно и не одной транзакцией с
- вставкой; при нескольких API-репликах возможен небольшой перебор.
-- Время восстановления backpressure включает `docker compose start` и
- healthcheck, поэтому это верхняя оценка, а не время чистой обработки.
+- `exactly_right` кодирует 1.64 MP WebP, `huge_few` после downscale — ~1.23 MP:
+ сценарий убирает scaling, но не уравнивает объём encode.
+- PNG-фикстуры синтетические; один прогон на ячейку, p95 по 12–18 заданиям
+ чувствителен к выбросу.
+- Время recovery включает `docker compose start` и healthcheck.
## Ограничения и следующие шаги
diff --git a/evidence/track1.html b/evidence/track1.html
index 53e7638b..ada26941 100644
--- a/evidence/track1.html
+++ b/evidence/track1.html
@@ -49,18 +49,18 @@ code { background: #f1f5f9; padding: 1px 5px; border-radius: 4px; }
Lab5, трек 1: очередь под перегрузкой
-Сгенерировано 2026-11-14T21:42:03.122104+00:00. Стенд: http://172.18.0.1:8080,
+
Сгенерировано 2026-11-20T16:02:07.856909+00:00. Стенд: http://172.18.0.1:8080,
4 CPU, Docker 5.6.1.
-Гипотеза. Маленькие изображения, которые почти не требуют
-масштабирования, должны давать высокий throughput и низкий RSS на задание.
-Большие изображения требуют декодирования и downscale, поэтому каждое задание
-дороже: при том же числе worker-ов очередь растёт, p95 time-to-ready
-увеличивается, а пиковый RSS зависит от concurrency сильнее, чем у маленьких
-файлов. Одновременно проверяем компромисс: eager уменьшает ожидание,
-но повышает потребление памяти; serial экономит память, но ограничивает
-throughput.
+Гипотеза. Стоимость задания складывается из decode, масштабирования
+и сжатия. exactly_right (1280×1280) не масштабирует optimized-вариант,
+поэтому показывает нижнюю границу стоимости сжатия; huge_few добавляет
+downscale, а small_many почти не масштабируется, но и пикселей в нём
+мало. Большие изображения при том же числе worker-ов увеличивают очередь,
+p95 time-to-ready и RSS. Одновременно проверяем компромисс: eager
+уменьшает ожидание, но повышает память; serial экономит память, но
+ограничивает throughput.
@@ -72,16 +72,15 @@ throughput.
Сценарии:
-small_many — 192×128, 30 jobs, many small images; variants fit or crop without large scaling; huge_few — 2400×1800, 12 jobs, few 4.32 MP images; optimized variant must scale down to 1280.
+small_many — 192×128, 30 jobs, many small images; almost no scaling; exactly_right — 1280×1280, 18 jobs, largest variant encodes at native 1280, so no scaling on the WebP path; huge_few — 2400×1800, 12 jobs, few 4.32 MP images; optimized variant must scale down to 1280.
Ключевые числа
-
-
1122 MiB
максимальный RSS worker
-
-
7.6 s
восстановление после backpressure
+
+
950 MiB
максимальный RSS worker
+
7.7 s
восстановление после backpressure
@@ -91,27 +90,34 @@ small_many — 192×128, 30 jobs, many small images; variants fit or crop withou
0 jobs/s
-4 jobs/s
+5 jobs/s
-8 jobs/s
+9 jobs/s
-12 jobs/s
+14 jobs/s
-16 jobs/s
-
-6.88 jobs/s
-
-11.59 jobs/s
-
-13.86 jobs/s
-small_many
-
-0.44 jobs/s
-
-0.45 jobs/s
-
-0.46 jobs/s
-huge_few
+18 jobs/s
+
+6.84 jobs/s
+
+13.69 jobs/s
+
+15.58 jobs/s
+small_many
+
+0.47 jobs/s
+
+0.49 jobs/s
+
+0.49 jobs/s
+exactly_right
+
+0.44 jobs/s
+
+0.44 jobs/s
+
+0.44 jobs/s
+huge_few
serial
@@ -126,27 +132,34 @@ small_many — 192×128, 30 jobs, many small images; variants fit or crop withou
0 ms
-7606 ms
+10560 ms
-15213 ms
+21120 ms
-22819 ms
+31680 ms
-30426 ms
-
-3284 ms
-
-1736 ms
-
-1158 ms
-small_many
-
-25521 ms
-
-25785 ms
-
-25699 ms
-huge_few
+42240 ms
+
+3338 ms
+
+1274 ms
+
+1166 ms
+small_many
+
+35714 ms
+
+35764 ms
+
+35797 ms
+exactly_right
+
+25896 ms
+
+26576 ms
+
+26641 ms
+huge_few
serial
@@ -161,27 +174,34 @@ small_many — 192×128, 30 jobs, many small images; variants fit or crop withou
0 ms
-6883 ms
+9909 ms
-13766 ms
+19819 ms
-20650 ms
+29728 ms
-27533 ms
-
-3194 ms
-
-1624 ms
-
-898 ms
-small_many
-
-23333 ms
-
-17716 ms
-
-17538 ms
-huge_few
+39637 ms
+
+3256 ms
+
+1147 ms
+
+924 ms
+small_many
+
+33591 ms
+
+31510 ms
+
+31735 ms
+exactly_right
+
+23668 ms
+
+18161 ms
+
+17648 ms
+huge_few
serial
@@ -196,27 +216,76 @@ small_many — 192×128, 30 jobs, many small images; variants fit or crop withou
0 MiB
-331 MiB
+280 MiB
-662 MiB
+560 MiB
-993 MiB
+840 MiB
-1324 MiB
-
-28 MiB
-
-31 MiB
-
-33 MiB
-small_many
-
-205 MiB
-
-729 MiB
-
-1122 MiB
-huge_few
+1121 MiB
+
+28 MiB
+
+29 MiB
+
+32 MiB
+small_many
+
+174 MiB
+
+434 MiB
+
+818 MiB
+exactly_right
+
+213 MiB
+
+756 MiB
+
+950 MiB
+huge_few
+
+serial
+
+balanced
+
+eager
+lower is better
+
+
+Compression overhead: доля encode в pipeline CPU
+encode / (decode + scale + encode), % (меньше — дешевле сжатие).
+
+0 %
+
+25 %
+
+49 %
+
+74 %
+
+99 %
+
+83.8 %
+
+83.8 %
+
+79.9 %
+small_many
+
+82.0 %
+
+83.1 %
+
+83.1 %
+exactly_right
+
+63.8 %
+
+61.4 %
+
+61.4 %
+huge_few
serial
@@ -229,11 +298,16 @@ small_many — 192×128, 30 jobs, many small images; variants fit or crop withou
Полная матрица
-Профиль Сценарий Jobs Jobs/s Queue lag p95, ms Ready p95, ms Processing p95, ms RSS peak, MiB Max parallel Max backlog
-serial small_many 30 6.88 3194 3284 152 28 1 27 serial huge_few 12 0.44 23333 25521 2311 205 1 12 balanced small_many 30 11.59 1624 1736 336 31 4 26 balanced huge_few 12 0.45 17716 25785 8710 729 4 12 eager small_many 30 13.86 898 1158 566 33 8 26 eager huge_few 12 0.46 17538 25699 17320 1122 8 12
+Профиль Сценарий Jobs Jobs/s Queue lag p95, ms Ready p95, ms Processing p95, ms Encode p95, ms Encode share, % RSS peak, MiB Max parallel Max backlog
+serial small_many 30 6.84 3256 3338 170 80 83.8 28 1 27 serial exactly_right 18 0.47 33591 35714 2180 1722 82.0 174 1 18 serial huge_few 12 0.44 23668 25896 2359 1508 63.8 213 1 12 balanced small_many 30 13.69 1147 1274 396 90 83.8 29 4 24 balanced exactly_right 18 0.49 31510 35764 8174 6562 83.1 434 4 18 balanced huge_few 12 0.44 18161 26576 9046 5606 61.4 756 4 12 eager small_many 30 15.58 924 1166 410 95 79.9 32 8 22 eager exactly_right 18 0.49 31735 35797 16363 13330 83.1 818 8 18 eager huge_few 12 0.44 17648 26641 17678 11124 61.4 950 8 12
Queue lag = processing_started_at - created_at.
Time-to-ready = finished_at - created_at. Processing = finished_at - processing_started_at.
+Encode = JPEG thumbnail + JPEG crop + WebP optimized; encode share =
+encode / (decode + scale + encode) из worker-лога.
+Суммы байт вариантов и оригинала остаются в JSON, но не выносятся в заголовок:
+синтетический PNG сжимается слишком хорошо, чтобы отношение байт было честной
+метрикой эффективности кодека.
RSS — VmRSS процесса worker, сэмплирование примерно каждые 100 мс.
Max parallel — максимум одновременно обрабатываемых заданий; max backlog —
максимум принятых, но ещё не готовых заданий.
@@ -242,20 +316,22 @@ Max parallel — максимум одновременно обрабатыва
Backpressure и восстановление
Перед этим прогоном API пересоздан с MAX_PENDING_JOBS=3, worker
-остановлен. Последовательные upload-ы показывают, где заканчивается принятая
-работа. Затем worker запускается, принятые задания должны дойти до ready.
+остановлен. Cap выбран намеренно: важно не число принятых заданий, а то, что
+API возвращает 503 с Retry-After и не теряет уже
+подтверждённую работу. Затем worker запускается, принятые задания должны дойти
+до ready.
# HTTP Latency, ms Retry-After Error code
-1 202 18.2 — — 2 202 17.4 — — 3 202 10.8 — — 4 503 3.9 5 unavailable
+1 202 26.7 — — 2 202 20.3 — — 3 202 13.8 — — 4 503 2.4 5 unavailable
Recovery: 3 принятых заданий
-стали ready за 7.60 с после старта worker.
+стали ready за 7.65 с после старта worker.
Финальный probe upload: HTTP 202 → ready.
Что это значит
-small_many: eager быстрее serial в 2.0× по throughput. small_many: пиковый RSS eager относится к serial как 1.2×. huge_few: throughput eager и serial практически одинаков. huge_few: пиковый RSS eager относится к serial как 5.5×. При balanced p95 time-to-ready для huge_few в 14.9× больше, чем для small_many: размер изображения меняет стоимость задания, а не только число заданий. На huge_few eager почти не поднял throughput (1.04×): worker ограничен 1.0 CPU, а декодирование и масштабирование упираются в CPU. Concurrency снижает queue lag, но при этом растёт RSS; это и есть компромисс latency/ресурсы, а не бесплатное ускорение. serial ограничивает одновременно обрабатываемые задания одним, поэтому queue lag растёт, но RSS остаётся минимальным. eager держит больше заданий в работе и в prefetch-окне брокера. Backpressure: при MAX_PENDING_JOBS=3 API принял 3 задания, затем вернул 503 с Retry-After=5; после старта worker очередь восстановилась за 7.6 с.
+small_many: eager быстрее serial в 2.3× по throughput. small_many: пиковый RSS eager относится к serial как 1.1×. exactly_right: throughput eager и serial практически одинаков. exactly_right: пиковый RSS eager относится к serial как 4.7×. huge_few: throughput eager и serial практически одинаков. huge_few: пиковый RSS eager относится к serial как 4.5×. При balanced p95 time-to-ready для huge_few в 20.9× больше, чем для small_many: размер изображения меняет стоимость задания, а не только число заданий. Compression vs scaling (balanced): exactly_right — encode share 83.1%, scale p95 1527 ms; huge_few — encode share 61.4%, scale p95 3448 ms. На exactly_right optimized WebP кодируется в native 1280, поэтому это нижняя граница стоимости сжатия; на huge_few к ней добавляется downscale. На huge_few eager почти не поднял throughput (1.02×): worker ограничен 1.0 CPU, а декодирование и масштабирование упираются в CPU. Concurrency снижает queue lag, но при этом растёт RSS; это и есть компромисс latency/ресурсы, а не бесплатное ускорение. serial ограничивает одновременно обрабатываемые задания одним, поэтому queue lag растёт, но RSS остаётся минимальным. eager держит больше заданий в работе и в prefetch-окне брокера. Backpressure: cap=3 выставлен намеренно, поэтому число принятых заданий — следствие настройки, а не открытие. Значимо то, что API вернул 503 с Retry-After=5, а после старта worker очередь восстановилась за 7.7 с.
@@ -263,8 +339,15 @@ Max parallel — максимум одновременно обрабатыва
Один worker, один RabbitMQ и один PostgreSQL: это сравнение настроек, а не
HA-кластер. Числа зависят от CPU/диска конкретной машины.
-PNG-фикстуры синтетические и хорошо сжимаются; доминирует стоимость декодирования
-и масштабирования, а не передача по сети. Размер в байтах указан в JSON.
+PNG-фикстуры синтетические и хорошо сжимаются; absolute encode time зависит
+от паттерна, но сравнение сценариев при одном паттерне остаётся корректным.
+Размер в байтах указан в JSON.
+Timing-поля читаются из worker-лога. Это wall-clock внутри процесса, поэтому
+при concurrency > 1 в них входит и планировщик, что как раз и нужно для
+сравнения профилей.
+exactly_right кодирует 1.64 MP WebP, а huge_few после downscale — около
+1.23 MP. Сценарий убирает scaling, но не уравнивает объём encode; это указано,
+чтобы не сравнивать throughput как «одинаковую работу».
RSS сэмплируется из /proc/1/status; короткий прогон не заменяет
длительный memory profile и p99.
Backpressure cap приблизительный: каждая API-реплика считает pending-задания
diff --git a/evidence/track1.json b/evidence/track1.json
index d9adae62..81fba6b0 100644
--- a/evidence/track1.json
+++ b/evidence/track1.json
@@ -1,8 +1,8 @@
{
- "generated_at": "2026-11-14T21:42:03.122104+00:00",
+ "generated_at": "2026-11-20T16:02:07.856909+00:00",
"environment": {
"base_url": "http://172.18.0.1:8080",
- "host": "48f08d2e9267",
+ "host": "8d2280b0dfc2",
"platform": "Linux-6.12.73+deb13-amd64-x86_64-with-glibc2.39",
"cpu_count": 4,
"python": "3.12.3",
@@ -15,7 +15,7 @@
"worker_mem_limit_default": "1G",
"note": "Experimental workers use a larger memory limit than .env so the eager huge-image run can hold decoded images without OOM; CPU is still limited so throughput is not a host-CPU artifact."
},
- "hypothesis": "Small images that fit the variants without large scaling should be cheaper per job; huge images force decode+downscale, so each job costs more memory and time. More concurrency should raise throughput and RSS but reduce queue lag.",
+ "hypothesis": "Job cost is decode + scale + encode. exactly_right (1280x1280) keeps the optimized WebP at native size, so it measures the compression floor; huge_few adds downscale, and small_many has few pixels. More concurrency should raise throughput and RSS but reduce queue lag.",
"profiles": [
{
"name": "serial",
@@ -41,10 +41,18 @@
"name": "small_many",
"width": 192,
"height": 128,
- "description": "many small images; variants fit or crop without large scaling",
+ "description": "many small images; almost no scaling",
"count": 30,
"bytes": 941
},
+ {
+ "name": "exactly_right",
+ "width": 1280,
+ "height": 1280,
+ "description": "largest variant encodes at native 1280, so no scaling on the WebP path",
+ "count": 18,
+ "bytes": 23176
+ },
{
"name": "huge_few",
"width": 2400,
@@ -65,72 +73,94 @@
"image_height": 128,
"image_bytes": 941,
"image_sha256": "7e90807e4df32b24ab7e0f8f540c76b93ed2a0e9f91e7b3350748cfc8029a088",
- "throughput_jobs_per_s": 6.875,
- "pipeline_span_seconds": 4.364,
- "submit_seconds": 0.987,
- "wall_seconds": 4.922,
- "upload_p50_ms": 43.4,
- "upload_p95_ms": 99.8,
- "queue_lag_p50_ms": 2030.8,
- "queue_lag_p95_ms": 3193.7,
- "queue_lag_max_ms": 3385.0,
- "time_to_ready_p50_ms": 2119.7,
- "time_to_ready_p95_ms": 3284.1,
- "time_to_ready_max_ms": 3475.8,
- "processing_p50_ms": 85.7,
- "processing_p95_ms": 151.5,
+ "throughput_jobs_per_s": 6.837,
+ "pipeline_span_seconds": 4.388,
+ "submit_seconds": 1.051,
+ "wall_seconds": 4.99,
+ "upload_p50_ms": 67.7,
+ "upload_p95_ms": 148.2,
+ "queue_lag_p50_ms": 2029.3,
+ "queue_lag_p95_ms": 3256.2,
+ "queue_lag_max_ms": 3425.7,
+ "time_to_ready_p50_ms": 2112.6,
+ "time_to_ready_p95_ms": 3338.0,
+ "time_to_ready_max_ms": 3506.6,
+ "processing_p50_ms": 88.4,
+ "processing_p95_ms": 169.9,
+ "decode_ms_p50": 0.26,
+ "decode_ms_p95": 1.52,
+ "scale_ms_p50": 5.34,
+ "scale_ms_p95": 14.31,
+ "jpeg_encode_ms_p50": 2.37,
+ "jpeg_encode_ms_p95": 5.96,
+ "webp_encode_ms_p50": 28.57,
+ "webp_encode_ms_p95": 75.86,
+ "encode_ms_p50": 30.71,
+ "encode_ms_p95": 80.0,
+ "pipeline_ms_p50": 36.3,
+ "pipeline_ms_p95": 92.23,
+ "encode_share_pct": 83.8,
+ "timing_jobs": 30,
+ "original_bytes": 28230,
+ "variants_bytes": 469710,
+ "variant_bytes": {
+ "thumbnail": 214740,
+ "crop": 151290,
+ "optimized": 103680
+ },
+ "compression_ratio": 16.639,
"max_backlog_jobs": 27,
"max_parallel_jobs": 1,
- "rss_baseline_mib": 20.3,
+ "rss_baseline_mib": 20.4,
"rss_peak_mib": 28.2,
- "rss_end_mib": 27.1,
+ "rss_end_mib": 27.6,
"rss_samples": 63,
- "rss_delta_mib": 7.9,
+ "rss_delta_mib": 7.8,
"timeline": [
{
- "t": 1.356,
+ "t": 1.408,
"queued": 24,
- "processing": 0,
- "ready": 6,
- "failed": 0
- },
- {
- "t": 1.998,
- "queued": 19,
"processing": 1,
- "ready": 10,
+ "ready": 5,
"failed": 0
},
{
- "t": 2.711,
- "queued": 15,
+ "t": 2.169,
+ "queued": 19,
"processing": 0,
+ "ready": 11,
+ "failed": 0
+ },
+ {
+ "t": 2.78,
+ "queued": 14,
+ "processing": 1,
"ready": 15,
"failed": 0
},
{
- "t": 3.265,
+ "t": 3.314,
"queued": 10,
"processing": 1,
"ready": 19,
"failed": 0
},
{
- "t": 3.754,
+ "t": 3.848,
"queued": 5,
- "processing": 1,
- "ready": 24,
+ "processing": 0,
+ "ready": 25,
"failed": 0
},
{
- "t": 4.384,
- "queued": 0,
- "processing": 1,
+ "t": 4.465,
+ "queued": 1,
+ "processing": 0,
"ready": 29,
"failed": 0
},
{
- "t": 4.922,
+ "t": 4.99,
"queued": 0,
"processing": 0,
"ready": 30,
@@ -138,6 +168,517 @@
}
]
},
+ {
+ "profile": "serial",
+ "scenario": "exactly_right",
+ "concurrency": 1,
+ "prefetch": 1,
+ "jobs": 18,
+ "image_width": 1280,
+ "image_height": 1280,
+ "image_bytes": 23176,
+ "image_sha256": "031edc2b1c3dc17d244613b1488661ffb44f5a70d425c0ec753287c3d5c41bdb",
+ "throughput_jobs_per_s": 0.475,
+ "pipeline_span_seconds": 37.868,
+ "submit_seconds": 0.7,
+ "wall_seconds": 38.475,
+ "upload_p50_ms": 63.0,
+ "upload_p95_ms": 130.9,
+ "queue_lag_p50_ms": 17769.1,
+ "queue_lag_p95_ms": 33590.9,
+ "queue_lag_max_ms": 35435.2,
+ "time_to_ready_p50_ms": 19857.3,
+ "time_to_ready_p95_ms": 35713.9,
+ "time_to_ready_max_ms": 37398.5,
+ "processing_p50_ms": 2034.6,
+ "processing_p95_ms": 2179.6,
+ "decode_ms_p50": 13.95,
+ "decode_ms_p95": 20.18,
+ "scale_ms_p50": 331.73,
+ "scale_ms_p95": 432.28,
+ "jpeg_encode_ms_p50": 4.58,
+ "jpeg_encode_ms_p95": 5.56,
+ "webp_encode_ms_p50": 1613.4,
+ "webp_encode_ms_p95": 1717.25,
+ "encode_ms_p50": 1617.88,
+ "encode_ms_p95": 1722.26,
+ "pipeline_ms_p50": 1976.81,
+ "pipeline_ms_p95": 2117.42,
+ "encode_share_pct": 82.0,
+ "timing_jobs": 18,
+ "original_bytes": 417168,
+ "variants_bytes": 4345668,
+ "variant_bytes": {
+ "thumbnail": 565254,
+ "crop": 142146,
+ "optimized": 3638268
+ },
+ "compression_ratio": 10.417,
+ "max_backlog_jobs": 18,
+ "max_parallel_jobs": 1,
+ "rss_baseline_mib": 20.5,
+ "rss_peak_mib": 174.3,
+ "rss_end_mib": 147.9,
+ "rss_samples": 382,
+ "rss_delta_mib": 153.8,
+ "timeline": [
+ {
+ "t": 1.084,
+ "queued": 17,
+ "processing": 1,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 1.569,
+ "queued": 17,
+ "processing": 1,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 2.023,
+ "queued": 17,
+ "processing": 1,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 2.824,
+ "queued": 16,
+ "processing": 1,
+ "ready": 1,
+ "failed": 0
+ },
+ {
+ "t": 3.53,
+ "queued": 16,
+ "processing": 1,
+ "ready": 1,
+ "failed": 0
+ },
+ {
+ "t": 3.993,
+ "queued": 16,
+ "processing": 1,
+ "ready": 1,
+ "failed": 0
+ },
+ {
+ "t": 4.63,
+ "queued": 16,
+ "processing": 1,
+ "ready": 1,
+ "failed": 0
+ },
+ {
+ "t": 5.174,
+ "queued": 15,
+ "processing": 1,
+ "ready": 2,
+ "failed": 0
+ },
+ {
+ "t": 5.651,
+ "queued": 15,
+ "processing": 1,
+ "ready": 2,
+ "failed": 0
+ },
+ {
+ "t": 6.189,
+ "queued": 15,
+ "processing": 1,
+ "ready": 2,
+ "failed": 0
+ },
+ {
+ "t": 6.79,
+ "queued": 15,
+ "processing": 0,
+ "ready": 3,
+ "failed": 0
+ },
+ {
+ "t": 7.493,
+ "queued": 14,
+ "processing": 1,
+ "ready": 3,
+ "failed": 0
+ },
+ {
+ "t": 8.013,
+ "queued": 14,
+ "processing": 1,
+ "ready": 3,
+ "failed": 0
+ },
+ {
+ "t": 8.619,
+ "queued": 14,
+ "processing": 1,
+ "ready": 3,
+ "failed": 0
+ },
+ {
+ "t": 9.193,
+ "queued": 13,
+ "processing": 1,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 9.655,
+ "queued": 13,
+ "processing": 1,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 10.192,
+ "queued": 13,
+ "processing": 1,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 10.798,
+ "queued": 13,
+ "processing": 1,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 11.314,
+ "queued": 12,
+ "processing": 1,
+ "ready": 5,
+ "failed": 0
+ },
+ {
+ "t": 11.845,
+ "queued": 12,
+ "processing": 1,
+ "ready": 5,
+ "failed": 0
+ },
+ {
+ "t": 12.651,
+ "queued": 12,
+ "processing": 1,
+ "ready": 5,
+ "failed": 0
+ },
+ {
+ "t": 13.251,
+ "queued": 11,
+ "processing": 1,
+ "ready": 6,
+ "failed": 0
+ },
+ {
+ "t": 13.738,
+ "queued": 11,
+ "processing": 1,
+ "ready": 6,
+ "failed": 0
+ },
+ {
+ "t": 14.334,
+ "queued": 11,
+ "processing": 1,
+ "ready": 6,
+ "failed": 0
+ },
+ {
+ "t": 14.971,
+ "queued": 11,
+ "processing": 1,
+ "ready": 6,
+ "failed": 0
+ },
+ {
+ "t": 15.461,
+ "queued": 10,
+ "processing": 1,
+ "ready": 7,
+ "failed": 0
+ },
+ {
+ "t": 15.968,
+ "queued": 10,
+ "processing": 1,
+ "ready": 7,
+ "failed": 0
+ },
+ {
+ "t": 16.806,
+ "queued": 10,
+ "processing": 1,
+ "ready": 7,
+ "failed": 0
+ },
+ {
+ "t": 17.396,
+ "queued": 9,
+ "processing": 1,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 17.865,
+ "queued": 9,
+ "processing": 1,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 18.498,
+ "queued": 9,
+ "processing": 1,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 19.066,
+ "queued": 9,
+ "processing": 1,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 19.551,
+ "queued": 8,
+ "processing": 1,
+ "ready": 9,
+ "failed": 0
+ },
+ {
+ "t": 20.02,
+ "queued": 8,
+ "processing": 1,
+ "ready": 9,
+ "failed": 0
+ },
+ {
+ "t": 20.72,
+ "queued": 8,
+ "processing": 1,
+ "ready": 9,
+ "failed": 0
+ },
+ {
+ "t": 21.467,
+ "queued": 7,
+ "processing": 1,
+ "ready": 10,
+ "failed": 0
+ },
+ {
+ "t": 21.942,
+ "queued": 7,
+ "processing": 1,
+ "ready": 10,
+ "failed": 0
+ },
+ {
+ "t": 22.584,
+ "queued": 7,
+ "processing": 1,
+ "ready": 10,
+ "failed": 0
+ },
+ {
+ "t": 23.161,
+ "queued": 7,
+ "processing": 1,
+ "ready": 10,
+ "failed": 0
+ },
+ {
+ "t": 23.639,
+ "queued": 6,
+ "processing": 1,
+ "ready": 11,
+ "failed": 0
+ },
+ {
+ "t": 24.135,
+ "queued": 6,
+ "processing": 1,
+ "ready": 11,
+ "failed": 0
+ },
+ {
+ "t": 24.834,
+ "queued": 6,
+ "processing": 1,
+ "ready": 11,
+ "failed": 0
+ },
+ {
+ "t": 25.533,
+ "queued": 6,
+ "processing": 1,
+ "ready": 11,
+ "failed": 0
+ },
+ {
+ "t": 26.068,
+ "queued": 5,
+ "processing": 1,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 26.675,
+ "queued": 5,
+ "processing": 1,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 27.25,
+ "queued": 5,
+ "processing": 1,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 27.893,
+ "queued": 4,
+ "processing": 1,
+ "ready": 13,
+ "failed": 0
+ },
+ {
+ "t": 28.359,
+ "queued": 4,
+ "processing": 1,
+ "ready": 13,
+ "failed": 0
+ },
+ {
+ "t": 28.83,
+ "queued": 4,
+ "processing": 1,
+ "ready": 13,
+ "failed": 0
+ },
+ {
+ "t": 29.637,
+ "queued": 4,
+ "processing": 0,
+ "ready": 14,
+ "failed": 0
+ },
+ {
+ "t": 30.377,
+ "queued": 3,
+ "processing": 1,
+ "ready": 14,
+ "failed": 0
+ },
+ {
+ "t": 30.885,
+ "queued": 3,
+ "processing": 1,
+ "ready": 14,
+ "failed": 0
+ },
+ {
+ "t": 31.533,
+ "queued": 3,
+ "processing": 1,
+ "ready": 14,
+ "failed": 0
+ },
+ {
+ "t": 32.214,
+ "queued": 2,
+ "processing": 1,
+ "ready": 15,
+ "failed": 0
+ },
+ {
+ "t": 32.68,
+ "queued": 2,
+ "processing": 1,
+ "ready": 15,
+ "failed": 0
+ },
+ {
+ "t": 33.236,
+ "queued": 2,
+ "processing": 1,
+ "ready": 15,
+ "failed": 0
+ },
+ {
+ "t": 33.951,
+ "queued": 1,
+ "processing": 1,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 34.599,
+ "queued": 1,
+ "processing": 1,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 35.142,
+ "queued": 1,
+ "processing": 1,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 35.776,
+ "queued": 1,
+ "processing": 1,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 36.27,
+ "queued": 0,
+ "processing": 1,
+ "ready": 17,
+ "failed": 0
+ },
+ {
+ "t": 36.763,
+ "queued": 0,
+ "processing": 1,
+ "ready": 17,
+ "failed": 0
+ },
+ {
+ "t": 37.393,
+ "queued": 0,
+ "processing": 1,
+ "ready": 17,
+ "failed": 0
+ },
+ {
+ "t": 38.0,
+ "queued": 0,
+ "processing": 1,
+ "ready": 17,
+ "failed": 0
+ },
+ {
+ "t": 38.475,
+ "queued": 0,
+ "processing": 0,
+ "ready": 18,
+ "failed": 0
+ }
+ ]
+ },
{
"profile": "serial",
"scenario": "huge_few",
@@ -148,352 +689,367 @@
"image_height": 1800,
"image_bytes": 53135,
"image_sha256": "68c88fcbe5c4ba31ee745047e8be4dc64e5d7711f81f5e50a4276fb52cf925ba",
- "throughput_jobs_per_s": 0.443,
- "pipeline_span_seconds": 27.106,
- "submit_seconds": 0.52,
- "wall_seconds": 27.417,
- "upload_p50_ms": 162.6,
- "upload_p95_ms": 227.5,
- "queue_lag_p50_ms": 12336.3,
- "queue_lag_p95_ms": 23333.0,
- "queue_lag_max_ms": 24511.6,
- "time_to_ready_p50_ms": 14530.2,
- "time_to_ready_p95_ms": 25521.0,
- "time_to_ready_max_ms": 26776.2,
- "processing_p50_ms": 2233.0,
- "processing_p95_ms": 2311.1,
+ "throughput_jobs_per_s": 0.436,
+ "pipeline_span_seconds": 27.539,
+ "submit_seconds": 0.724,
+ "wall_seconds": 27.861,
+ "upload_p50_ms": 220.5,
+ "upload_p95_ms": 349.6,
+ "queue_lag_p50_ms": 12568.0,
+ "queue_lag_p95_ms": 23668.1,
+ "queue_lag_max_ms": 24910.1,
+ "time_to_ready_p50_ms": 14822.7,
+ "time_to_ready_p95_ms": 25896.4,
+ "time_to_ready_max_ms": 27051.7,
+ "processing_p50_ms": 2213.4,
+ "processing_p95_ms": 2359.1,
+ "decode_ms_p50": 36.44,
+ "decode_ms_p95": 43.65,
+ "scale_ms_p50": 751.12,
+ "scale_ms_p95": 836.9,
+ "jpeg_encode_ms_p50": 3.46,
+ "jpeg_encode_ms_p95": 4.62,
+ "webp_encode_ms_p50": 1366.74,
+ "webp_encode_ms_p95": 1504.85,
+ "encode_ms_p50": 1370.34,
+ "encode_ms_p95": 1508.26,
+ "pipeline_ms_p50": 2148.57,
+ "pipeline_ms_p95": 2287.69,
+ "encode_share_pct": 63.8,
+ "timing_jobs": 12,
+ "original_bytes": 637620,
+ "variants_bytes": 3163980,
+ "variant_bytes": {
+ "thumbnail": 260148,
+ "crop": 78672,
+ "optimized": 2825160
+ },
+ "compression_ratio": 4.962,
"max_backlog_jobs": 12,
"max_parallel_jobs": 1,
- "rss_baseline_mib": 20.5,
- "rss_peak_mib": 204.7,
- "rss_end_mib": 203.1,
- "rss_samples": 278,
- "rss_delta_mib": 184.2,
+ "rss_baseline_mib": 20.4,
+ "rss_peak_mib": 212.9,
+ "rss_end_mib": 201.9,
+ "rss_samples": 282,
+ "rss_delta_mib": 192.5,
"timeline": [
{
- "t": 0.827,
+ "t": 1.103,
"queued": 11,
"processing": 1,
"ready": 0,
"failed": 0
},
{
- "t": 1.494,
+ "t": 1.718,
"queued": 11,
"processing": 1,
"ready": 0,
"failed": 0
},
{
- "t": 2.002,
+ "t": 2.373,
"queued": 11,
"processing": 1,
"ready": 0,
"failed": 0
},
{
- "t": 2.488,
+ "t": 3.053,
"queued": 10,
"processing": 1,
"ready": 1,
"failed": 0
},
{
- "t": 3.193,
+ "t": 3.561,
"queued": 10,
"processing": 1,
"ready": 1,
"failed": 0
},
{
- "t": 4.011,
+ "t": 4.035,
"queued": 10,
"processing": 1,
"ready": 1,
"failed": 0
},
{
- "t": 4.519,
+ "t": 4.705,
"queued": 10,
"processing": 1,
"ready": 1,
"failed": 0
},
{
- "t": 5.153,
+ "t": 5.257,
"queued": 9,
"processing": 1,
"ready": 2,
"failed": 0
},
{
- "t": 5.723,
+ "t": 5.731,
"queued": 9,
"processing": 1,
"ready": 2,
"failed": 0
},
{
- "t": 6.187,
+ "t": 6.387,
"queued": 9,
"processing": 1,
"ready": 2,
"failed": 0
},
{
- "t": 6.671,
+ "t": 7.217,
"queued": 9,
"processing": 1,
"ready": 2,
"failed": 0
},
{
- "t": 7.283,
+ "t": 7.717,
"queued": 8,
"processing": 1,
"ready": 3,
"failed": 0
},
{
- "t": 7.94,
+ "t": 8.301,
"queued": 8,
"processing": 1,
"ready": 3,
"failed": 0
},
{
- "t": 8.452,
+ "t": 8.921,
"queued": 8,
"processing": 1,
"ready": 3,
"failed": 0
},
{
- "t": 9.103,
+ "t": 9.555,
"queued": 8,
"processing": 1,
"ready": 3,
"failed": 0
},
{
- "t": 9.694,
+ "t": 10.199,
"queued": 7,
"processing": 1,
"ready": 4,
"failed": 0
},
{
- "t": 10.185,
+ "t": 10.721,
"queued": 7,
"processing": 1,
"ready": 4,
"failed": 0
},
{
- "t": 10.674,
+ "t": 11.453,
"queued": 7,
"processing": 1,
"ready": 4,
"failed": 0
},
{
- "t": 11.333,
- "queued": 7,
- "processing": 1,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 11.836,
+ "t": 12.053,
"queued": 6,
"processing": 1,
"ready": 5,
"failed": 0
},
{
- "t": 12.386,
+ "t": 12.587,
"queued": 6,
"processing": 1,
"ready": 5,
"failed": 0
},
{
- "t": 13.112,
+ "t": 13.065,
"queued": 6,
"processing": 1,
"ready": 5,
"failed": 0
},
{
- "t": 13.692,
+ "t": 13.704,
"queued": 6,
"processing": 1,
"ready": 5,
"failed": 0
},
{
- "t": 14.188,
+ "t": 14.234,
"queued": 5,
"processing": 1,
"ready": 6,
"failed": 0
},
{
- "t": 14.684,
+ "t": 14.723,
"queued": 5,
"processing": 1,
"ready": 6,
"failed": 0
},
{
- "t": 15.277,
+ "t": 15.371,
"queued": 5,
"processing": 1,
"ready": 6,
"failed": 0
},
{
- "t": 15.82,
+ "t": 16.103,
"queued": 5,
"processing": 1,
"ready": 6,
"failed": 0
},
{
- "t": 16.477,
+ "t": 16.663,
"queued": 4,
"processing": 1,
"ready": 7,
"failed": 0
},
{
- "t": 17.103,
+ "t": 17.154,
"queued": 4,
"processing": 1,
"ready": 7,
"failed": 0
},
{
- "t": 17.622,
+ "t": 17.803,
"queued": 4,
"processing": 1,
"ready": 7,
"failed": 0
},
{
- "t": 18.203,
+ "t": 18.351,
"queued": 4,
"processing": 1,
"ready": 7,
"failed": 0
},
{
- "t": 18.738,
+ "t": 18.842,
"queued": 3,
"processing": 1,
"ready": 8,
"failed": 0
},
{
- "t": 19.242,
+ "t": 19.592,
"queued": 3,
"processing": 1,
"ready": 8,
"failed": 0
},
{
- "t": 19.806,
+ "t": 20.327,
"queued": 3,
"processing": 1,
"ready": 8,
"failed": 0
},
{
- "t": 20.406,
+ "t": 20.909,
"queued": 3,
"processing": 1,
"ready": 8,
"failed": 0
},
{
- "t": 20.89,
+ "t": 21.545,
"queued": 2,
"processing": 1,
"ready": 9,
"failed": 0
},
{
- "t": 21.44,
+ "t": 22.235,
"queued": 2,
"processing": 1,
"ready": 9,
"failed": 0
},
{
- "t": 22.162,
+ "t": 22.7,
"queued": 2,
"processing": 1,
"ready": 9,
"failed": 0
},
{
- "t": 22.773,
+ "t": 23.195,
"queued": 2,
"processing": 1,
"ready": 9,
"failed": 0
},
{
- "t": 23.257,
+ "t": 23.816,
"queued": 1,
"processing": 1,
"ready": 10,
"failed": 0
},
{
- "t": 23.854,
+ "t": 24.524,
"queued": 1,
"processing": 1,
"ready": 10,
"failed": 0
},
{
- "t": 24.452,
+ "t": 25.041,
"queued": 1,
"processing": 1,
"ready": 10,
"failed": 0
},
{
- "t": 24.927,
- "queued": 1,
- "processing": 1,
- "ready": 10,
- "failed": 0
- },
- {
- "t": 25.472,
+ "t": 25.696,
"queued": 0,
"processing": 1,
"ready": 11,
"failed": 0
},
{
- "t": 26.224,
+ "t": 26.264,
"queued": 0,
"processing": 1,
"ready": 11,
"failed": 0
},
{
- "t": 26.95,
+ "t": 26.733,
"queued": 0,
"processing": 1,
"ready": 11,
"failed": 0
},
{
- "t": 27.417,
+ "t": 27.275,
+ "queued": 0,
+ "processing": 1,
+ "ready": 11,
+ "failed": 0
+ },
+ {
+ "t": 27.861,
"queued": 0,
"processing": 0,
"ready": 12,
@@ -511,44 +1067,66 @@
"image_height": 128,
"image_bytes": 941,
"image_sha256": "7e90807e4df32b24ab7e0f8f540c76b93ed2a0e9f91e7b3350748cfc8029a088",
- "throughput_jobs_per_s": 11.59,
- "pipeline_span_seconds": 2.588,
- "submit_seconds": 1.049,
- "wall_seconds": 2.966,
- "upload_p50_ms": 58.0,
- "upload_p95_ms": 178.8,
- "queue_lag_p50_ms": 1131.4,
- "queue_lag_p95_ms": 1623.9,
- "queue_lag_max_ms": 1647.8,
- "time_to_ready_p50_ms": 1335.0,
- "time_to_ready_p95_ms": 1735.8,
- "time_to_ready_max_ms": 1756.1,
- "processing_p50_ms": 211.1,
- "processing_p95_ms": 336.3,
- "max_backlog_jobs": 26,
+ "throughput_jobs_per_s": 13.694,
+ "pipeline_span_seconds": 2.191,
+ "submit_seconds": 1.072,
+ "wall_seconds": 2.852,
+ "upload_p50_ms": 57.2,
+ "upload_p95_ms": 157.6,
+ "queue_lag_p50_ms": 916.5,
+ "queue_lag_p95_ms": 1146.6,
+ "queue_lag_max_ms": 1174.2,
+ "time_to_ready_p50_ms": 1042.9,
+ "time_to_ready_p95_ms": 1273.7,
+ "time_to_ready_max_ms": 1324.2,
+ "processing_p50_ms": 149.2,
+ "processing_p95_ms": 396.0,
+ "decode_ms_p50": 0.27,
+ "decode_ms_p95": 1.17,
+ "scale_ms_p50": 6.66,
+ "scale_ms_p95": 17.26,
+ "jpeg_encode_ms_p50": 2.4,
+ "jpeg_encode_ms_p95": 7.88,
+ "webp_encode_ms_p50": 35.93,
+ "webp_encode_ms_p95": 87.15,
+ "encode_ms_p50": 38.7,
+ "encode_ms_p95": 89.96,
+ "pipeline_ms_p50": 49.27,
+ "pipeline_ms_p95": 109.71,
+ "encode_share_pct": 83.8,
+ "timing_jobs": 30,
+ "original_bytes": 28230,
+ "variants_bytes": 469710,
+ "variant_bytes": {
+ "thumbnail": 214740,
+ "crop": 151290,
+ "optimized": 103680
+ },
+ "compression_ratio": 16.639,
+ "max_backlog_jobs": 24,
"max_parallel_jobs": 4,
- "rss_baseline_mib": 20.5,
- "rss_peak_mib": 30.6,
- "rss_end_mib": 28.5,
+ "rss_baseline_mib": 20.4,
+ "rss_peak_mib": 28.9,
+ "rss_end_mib": 27.4,
"rss_samples": 43,
- "rss_delta_mib": 10.1,
+ "rss_delta_mib": 8.5,
"timeline": [
{
- "t": 1.679,
- "queued": 14,
- "processing": 4,
+ "t": 1.489,
+ "queued": 15,
+ "processing": 3,
"ready": 12,
"failed": 0
},
{
- "t": 2.514,
- "queued": 4,
- "processing": 2,
- "ready": 24,
+ "t": 2.222,
+ "queued": 0,
+ "processing": 3,
+ "ready": 27,
"failed": 0
},
{
- "t": 2.966,
+ "t": 2.852,
"queued": 0,
"processing": 0,
"ready": 30,
@@ -558,373 +1136,900 @@
},
{
"profile": "balanced",
- "scenario": "huge_few",
+ "scenario": "exactly_right",
"concurrency": 4,
"prefetch": 4,
- "jobs": 12,
- "image_width": 2400,
- "image_height": 1800,
- "image_bytes": 53135,
- "image_sha256": "68c88fcbe5c4ba31ee745047e8be4dc64e5d7711f81f5e50a4276fb52cf925ba",
- "throughput_jobs_per_s": 0.454,
- "pipeline_span_seconds": 26.412,
- "submit_seconds": 0.688,
- "wall_seconds": 26.671,
- "upload_p50_ms": 178.3,
- "upload_p95_ms": 298.3,
- "queue_lag_p50_ms": 8712.6,
- "queue_lag_p95_ms": 17716.4,
- "queue_lag_max_ms": 18009.6,
- "time_to_ready_p50_ms": 17199.1,
- "time_to_ready_p95_ms": 25784.6,
- "time_to_ready_max_ms": 25919.7,
- "processing_p50_ms": 8459.9,
- "processing_p95_ms": 8709.7,
- "max_backlog_jobs": 12,
+ "jobs": 18,
+ "image_width": 1280,
+ "image_height": 1280,
+ "image_bytes": 23176,
+ "image_sha256": "031edc2b1c3dc17d244613b1488661ffb44f5a70d425c0ec753287c3d5c41bdb",
+ "throughput_jobs_per_s": 0.495,
+ "pipeline_span_seconds": 36.391,
+ "submit_seconds": 0.745,
+ "wall_seconds": 36.739,
+ "upload_p50_ms": 68.9,
+ "upload_p95_ms": 129.2,
+ "queue_lag_p50_ms": 15501.0,
+ "queue_lag_p95_ms": 31510.3,
+ "queue_lag_max_ms": 31697.3,
+ "time_to_ready_p50_ms": 23200.0,
+ "time_to_ready_p95_ms": 35764.4,
+ "time_to_ready_max_ms": 35844.6,
+ "processing_p50_ms": 7711.4,
+ "processing_p95_ms": 8173.6,
+ "decode_ms_p50": 21.65,
+ "decode_ms_p95": 85.9,
+ "scale_ms_p50": 1134.16,
+ "scale_ms_p95": 1527.0,
+ "jpeg_encode_ms_p50": 6.13,
+ "jpeg_encode_ms_p95": 35.49,
+ "webp_encode_ms_p50": 6165.41,
+ "webp_encode_ms_p95": 6554.13,
+ "encode_ms_p50": 6174.86,
+ "encode_ms_p95": 6562.25,
+ "pipeline_ms_p50": 7376.17,
+ "pipeline_ms_p95": 8011.08,
+ "encode_share_pct": 83.1,
+ "timing_jobs": 18,
+ "original_bytes": 417168,
+ "variants_bytes": 4345668,
+ "variant_bytes": {
+ "thumbnail": 565254,
+ "crop": 142146,
+ "optimized": 3638268
+ },
+ "compression_ratio": 10.417,
+ "max_backlog_jobs": 18,
"max_parallel_jobs": 4,
"rss_baseline_mib": 20.4,
- "rss_peak_mib": 728.7,
- "rss_end_mib": 609.9,
- "rss_samples": 254,
- "rss_delta_mib": 708.3,
+ "rss_peak_mib": 433.8,
+ "rss_end_mib": 234.6,
+ "rss_samples": 346,
+ "rss_delta_mib": 413.4,
"timeline": [
{
- "t": 1.028,
- "queued": 8,
+ "t": 1.145,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 1.694,
- "queued": 8,
+ "t": 1.641,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 2.319,
- "queued": 8,
+ "t": 2.489,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 2.905,
- "queued": 8,
+ "t": 3.204,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 3.537,
- "queued": 8,
+ "t": 3.668,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 4.173,
- "queued": 8,
+ "t": 4.342,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 4.829,
- "queued": 8,
+ "t": 4.881,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 5.508,
- "queued": 8,
+ "t": 5.367,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 6.11,
- "queued": 8,
+ "t": 5.967,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 6.603,
- "queued": 8,
+ "t": 6.779,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 7.14,
- "queued": 8,
+ "t": 7.316,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 7.796,
- "queued": 8,
+ "t": 7.926,
+ "queued": 14,
"processing": 4,
"ready": 0,
"failed": 0
},
{
- "t": 8.345,
- "queued": 8,
+ "t": 8.666,
+ "queued": 11,
"processing": 4,
- "ready": 0,
+ "ready": 3,
"failed": 0
},
{
- "t": 8.854,
- "queued": 8,
- "processing": 4,
- "ready": 0,
- "failed": 0
- },
- {
- "t": 9.667,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 10.488,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 11.103,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 11.769,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 12.384,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 12.857,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 13.371,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 14.228,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 14.929,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 15.574,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 16.234,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 16.711,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 17.257,
- "queued": 4,
- "processing": 4,
- "ready": 4,
- "failed": 0
- },
- {
- "t": 17.945,
- "queued": 2,
+ "t": 9.145,
+ "queued": 11,
"processing": 3,
- "ready": 7,
+ "ready": 4,
"failed": 0
},
{
- "t": 18.674,
- "queued": 1,
+ "t": 9.675,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 10.329,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 10.924,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 11.419,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 12.149,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 12.764,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 13.242,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 13.715,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 14.363,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 14.939,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 15.461,
+ "queued": 10,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 16.279,
+ "queued": 8,
+ "processing": 4,
+ "ready": 6,
+ "failed": 0
+ },
+ {
+ "t": 17.024,
+ "queued": 7,
"processing": 3,
"ready": 8,
"failed": 0
},
{
- "t": 19.251,
- "queued": 0,
+ "t": 17.515,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 19.997,
- "queued": 0,
+ "t": 18.146,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 20.662,
- "queued": 0,
+ "t": 18.745,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 21.146,
- "queued": 0,
+ "t": 19.224,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 21.758,
- "queued": 0,
+ "t": 19.778,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 22.395,
- "queued": 0,
+ "t": 20.561,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 22.945,
- "queued": 0,
+ "t": 21.335,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 23.693,
- "queued": 0,
+ "t": 22.006,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 24.381,
- "queued": 0,
+ "t": 22.619,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 24.906,
- "queued": 0,
+ "t": 23.122,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 25.597,
- "queued": 0,
+ "t": 23.625,
+ "queued": 6,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 26.174,
- "queued": 0,
- "processing": 4,
- "ready": 8,
+ "t": 24.239,
+ "queued": 5,
+ "processing": 3,
+ "ready": 10,
"failed": 0
},
{
- "t": 26.671,
- "queued": 0,
- "processing": 0,
+ "t": 25.008,
+ "queued": 4,
+ "processing": 2,
"ready": 12,
"failed": 0
- }
- ]
- },
- {
- "profile": "eager",
- "scenario": "small_many",
- "concurrency": 8,
- "prefetch": 32,
- "jobs": 30,
- "image_width": 192,
- "image_height": 128,
- "image_bytes": 941,
- "image_sha256": "7e90807e4df32b24ab7e0f8f540c76b93ed2a0e9f91e7b3350748cfc8029a088",
- "throughput_jobs_per_s": 13.858,
- "pipeline_span_seconds": 2.165,
- "submit_seconds": 1.168,
- "wall_seconds": 2.318,
- "upload_p50_ms": 46.4,
- "upload_p95_ms": 126.7,
- "queue_lag_p50_ms": 749.2,
- "queue_lag_p95_ms": 897.8,
- "queue_lag_max_ms": 1052.1,
- "time_to_ready_p50_ms": 1013.5,
- "time_to_ready_p95_ms": 1157.8,
- "time_to_ready_max_ms": 1175.0,
- "processing_p50_ms": 271.8,
- "processing_p95_ms": 565.9,
- "max_backlog_jobs": 26,
- "max_parallel_jobs": 8,
- "rss_baseline_mib": 20.4,
- "rss_peak_mib": 32.7,
- "rss_end_mib": 28.8,
- "rss_samples": 37,
- "rss_delta_mib": 12.3,
- "timeline": [
+ },
{
- "t": 1.637,
- "queued": 10,
- "processing": 6,
+ "t": 25.639,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 26.343,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 26.887,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 27.361,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 27.967,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 28.567,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 29.116,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 29.723,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 30.548,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 31.201,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 31.755,
+ "queued": 2,
+ "processing": 4,
+ "ready": 12,
+ "failed": 0
+ },
+ {
+ "t": 32.394,
+ "queued": 1,
+ "processing": 3,
"ready": 14,
"failed": 0
},
{
- "t": 2.318,
+ "t": 32.993,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 33.567,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 34.321,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 35.03,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 35.509,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 36.148,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 36.739,
+ "queued": 0,
+ "processing": 0,
+ "ready": 18,
+ "failed": 0
+ }
+ ]
+ },
+ {
+ "profile": "balanced",
+ "scenario": "huge_few",
+ "concurrency": 4,
+ "prefetch": 4,
+ "jobs": 12,
+ "image_width": 2400,
+ "image_height": 1800,
+ "image_bytes": 53135,
+ "image_sha256": "68c88fcbe5c4ba31ee745047e8be4dc64e5d7711f81f5e50a4276fb52cf925ba",
+ "throughput_jobs_per_s": 0.441,
+ "pipeline_span_seconds": 27.197,
+ "submit_seconds": 0.699,
+ "wall_seconds": 27.842,
+ "upload_p50_ms": 234.9,
+ "upload_p95_ms": 340.5,
+ "queue_lag_p50_ms": 9161.6,
+ "queue_lag_p95_ms": 18160.7,
+ "queue_lag_max_ms": 18615.8,
+ "time_to_ready_p50_ms": 17676.9,
+ "time_to_ready_p95_ms": 26575.8,
+ "time_to_ready_max_ms": 26687.1,
+ "processing_p50_ms": 8751.2,
+ "processing_p95_ms": 9046.1,
+ "decode_ms_p50": 75.2,
+ "decode_ms_p95": 173.97,
+ "scale_ms_p50": 3244.52,
+ "scale_ms_p95": 3447.76,
+ "jpeg_encode_ms_p50": 4.08,
+ "jpeg_encode_ms_p95": 72.28,
+ "webp_encode_ms_p50": 5224.87,
+ "webp_encode_ms_p95": 5601.71,
+ "encode_ms_p50": 5254.45,
+ "encode_ms_p95": 5605.8,
+ "pipeline_ms_p50": 8502.19,
+ "pipeline_ms_p95": 8893.1,
+ "encode_share_pct": 61.4,
+ "timing_jobs": 12,
+ "original_bytes": 637620,
+ "variants_bytes": 3163980,
+ "variant_bytes": {
+ "thumbnail": 260148,
+ "crop": 78672,
+ "optimized": 2825160
+ },
+ "compression_ratio": 4.962,
+ "max_backlog_jobs": 12,
+ "max_parallel_jobs": 4,
+ "rss_baseline_mib": 18.4,
+ "rss_peak_mib": 756.4,
+ "rss_end_mib": 700.1,
+ "rss_samples": 264,
+ "rss_delta_mib": 738.0,
+ "timeline": [
+ {
+ "t": 1.054,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 1.69,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 2.18,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 2.852,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 3.554,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 4.119,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 4.796,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 5.369,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 5.899,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 6.538,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 7.201,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 7.742,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 8.363,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 9.121,
+ "queued": 8,
+ "processing": 4,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 9.736,
+ "queued": 5,
+ "processing": 3,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 10.236,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 10.887,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 11.445,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 11.968,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 12.787,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 13.713,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 14.188,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 14.848,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 15.451,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 16.048,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 16.526,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 17.073,
+ "queued": 4,
+ "processing": 4,
+ "ready": 4,
+ "failed": 0
+ },
+ {
+ "t": 17.941,
+ "queued": 4,
+ "processing": 3,
+ "ready": 5,
+ "failed": 0
+ },
+ {
+ "t": 18.562,
+ "queued": 1,
+ "processing": 4,
+ "ready": 7,
+ "failed": 0
+ },
+ {
+ "t": 19.062,
+ "queued": 1,
+ "processing": 3,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 19.715,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 20.364,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 20.897,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 21.7,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 22.547,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 23.064,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 23.704,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 24.28,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 24.785,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 25.397,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 26.09,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 26.685,
+ "queued": 0,
+ "processing": 4,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 27.269,
+ "queued": 0,
+ "processing": 1,
+ "ready": 11,
+ "failed": 0
+ },
+ {
+ "t": 27.842,
+ "queued": 0,
+ "processing": 0,
+ "ready": 12,
+ "failed": 0
+ }
+ ]
+ },
+ {
+ "profile": "eager",
+ "scenario": "small_many",
+ "concurrency": 8,
+ "prefetch": 32,
+ "jobs": 30,
+ "image_width": 192,
+ "image_height": 128,
+ "image_bytes": 941,
+ "image_sha256": "7e90807e4df32b24ab7e0f8f540c76b93ed2a0e9f91e7b3350748cfc8029a088",
+ "throughput_jobs_per_s": 15.576,
+ "pipeline_span_seconds": 1.926,
+ "submit_seconds": 0.991,
+ "wall_seconds": 2.218,
+ "upload_p50_ms": 49.4,
+ "upload_p95_ms": 140.4,
+ "queue_lag_p50_ms": 620.0,
+ "queue_lag_p95_ms": 923.7,
+ "queue_lag_max_ms": 993.7,
+ "time_to_ready_p50_ms": 876.8,
+ "time_to_ready_p95_ms": 1165.8,
+ "time_to_ready_max_ms": 1192.1,
+ "processing_p50_ms": 288.9,
+ "processing_p95_ms": 410.2,
+ "decode_ms_p50": 0.28,
+ "decode_ms_p95": 2.92,
+ "scale_ms_p50": 11.84,
+ "scale_ms_p95": 39.22,
+ "jpeg_encode_ms_p50": 3.47,
+ "jpeg_encode_ms_p95": 25.67,
+ "webp_encode_ms_p50": 53.31,
+ "webp_encode_ms_p95": 88.79,
+ "encode_ms_p50": 60.16,
+ "encode_ms_p95": 94.53,
+ "pipeline_ms_p50": 70.72,
+ "pipeline_ms_p95": 116.55,
+ "encode_share_pct": 79.9,
+ "timing_jobs": 30,
+ "original_bytes": 28230,
+ "variants_bytes": 469710,
+ "variant_bytes": {
+ "thumbnail": 214740,
+ "crop": 151290,
+ "optimized": 103680
+ },
+ "compression_ratio": 16.639,
+ "max_backlog_jobs": 22,
+ "max_parallel_jobs": 8,
+ "rss_baseline_mib": 20.4,
+ "rss_peak_mib": 32.1,
+ "rss_end_mib": 31.3,
+ "rss_samples": 36,
+ "rss_delta_mib": 11.7,
+ "timeline": [
+ {
+ "t": 1.583,
+ "queued": 6,
+ "processing": 5,
+ "ready": 19,
+ "failed": 0
+ },
+ {
+ "t": 2.218,
"queued": 0,
"processing": 0,
"ready": 30,
@@ -932,6 +2037,482 @@
}
]
},
+ {
+ "profile": "eager",
+ "scenario": "exactly_right",
+ "concurrency": 8,
+ "prefetch": 32,
+ "jobs": 18,
+ "image_width": 1280,
+ "image_height": 1280,
+ "image_bytes": 23176,
+ "image_sha256": "031edc2b1c3dc17d244613b1488661ffb44f5a70d425c0ec753287c3d5c41bdb",
+ "throughput_jobs_per_s": 0.495,
+ "pipeline_span_seconds": 36.374,
+ "submit_seconds": 0.732,
+ "wall_seconds": 37.056,
+ "upload_p50_ms": 99.8,
+ "upload_p95_ms": 143.3,
+ "queue_lag_p50_ms": 16204.1,
+ "queue_lag_p95_ms": 31734.6,
+ "queue_lag_max_ms": 31738.1,
+ "time_to_ready_p50_ms": 31582.0,
+ "time_to_ready_p95_ms": 35797.0,
+ "time_to_ready_max_ms": 35874.2,
+ "processing_p50_ms": 15651.6,
+ "processing_p95_ms": 16362.7,
+ "decode_ms_p50": 15.19,
+ "decode_ms_p95": 79.65,
+ "scale_ms_p50": 2513.98,
+ "scale_ms_p95": 3215.38,
+ "jpeg_encode_ms_p50": 5.52,
+ "jpeg_encode_ms_p95": 110.2,
+ "webp_encode_ms_p50": 12503.33,
+ "webp_encode_ms_p95": 13300.65,
+ "encode_ms_p50": 12510.13,
+ "encode_ms_p95": 13330.36,
+ "pipeline_ms_p50": 15115.26,
+ "pipeline_ms_p95": 15839.92,
+ "encode_share_pct": 83.1,
+ "timing_jobs": 18,
+ "original_bytes": 417168,
+ "variants_bytes": 4345668,
+ "variant_bytes": {
+ "thumbnail": 565254,
+ "crop": 142146,
+ "optimized": 3638268
+ },
+ "compression_ratio": 10.417,
+ "max_backlog_jobs": 18,
+ "max_parallel_jobs": 8,
+ "rss_baseline_mib": 20.5,
+ "rss_peak_mib": 818.4,
+ "rss_end_mib": 267.5,
+ "rss_samples": 348,
+ "rss_delta_mib": 797.9,
+ "timeline": [
+ {
+ "t": 1.029,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 1.517,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 2.383,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 3.225,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 3.785,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 4.401,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 4.895,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 5.386,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 6.031,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 6.66,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 7.187,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 7.911,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 8.609,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 9.088,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 9.591,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 10.224,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 10.79,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 11.355,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 12.104,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 12.728,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 13.258,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 13.882,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 14.512,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 14.99,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 15.545,
+ "queued": 10,
+ "processing": 8,
+ "ready": 0,
+ "failed": 0
+ },
+ {
+ "t": 16.347,
+ "queued": 10,
+ "processing": 7,
+ "ready": 1,
+ "failed": 0
+ },
+ {
+ "t": 17.27,
+ "queued": 5,
+ "processing": 5,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 17.918,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 18.571,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 19.099,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 19.6,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 20.256,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 20.912,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 21.467,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 22.107,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 22.686,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 23.202,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 23.706,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 24.31,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 25.015,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 25.809,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 26.38,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 27.044,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 27.652,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 28.125,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 28.638,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 29.32,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 29.998,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 30.527,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 31.164,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 31.762,
+ "queued": 2,
+ "processing": 8,
+ "ready": 8,
+ "failed": 0
+ },
+ {
+ "t": 32.258,
+ "queued": 2,
+ "processing": 6,
+ "ready": 10,
+ "failed": 0
+ },
+ {
+ "t": 32.918,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 33.555,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 34.143,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 34.717,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 35.406,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 35.974,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 36.491,
+ "queued": 0,
+ "processing": 2,
+ "ready": 16,
+ "failed": 0
+ },
+ {
+ "t": 37.056,
+ "queued": 0,
+ "processing": 0,
+ "ready": 18,
+ "failed": 0
+ }
+ ]
+ },
{
"profile": "eager",
"scenario": "huge_few",
@@ -942,317 +2523,353 @@
"image_height": 1800,
"image_bytes": 53135,
"image_sha256": "68c88fcbe5c4ba31ee745047e8be4dc64e5d7711f81f5e50a4276fb52cf925ba",
- "throughput_jobs_per_s": 0.459,
- "pipeline_span_seconds": 26.164,
- "submit_seconds": 0.621,
- "wall_seconds": 26.514,
- "upload_p50_ms": 211.1,
- "upload_p95_ms": 334.8,
- "queue_lag_p50_ms": 603.5,
- "queue_lag_p95_ms": 17537.9,
- "queue_lag_max_ms": 17742.4,
- "time_to_ready_p50_ms": 17932.3,
- "time_to_ready_p95_ms": 25698.9,
- "time_to_ready_max_ms": 25761.3,
- "processing_p50_ms": 16723.3,
- "processing_p95_ms": 17319.7,
+ "throughput_jobs_per_s": 0.443,
+ "pipeline_span_seconds": 27.077,
+ "submit_seconds": 0.615,
+ "wall_seconds": 27.714,
+ "upload_p50_ms": 242.8,
+ "upload_p95_ms": 345.4,
+ "queue_lag_p50_ms": 464.0,
+ "queue_lag_p95_ms": 17648.0,
+ "queue_lag_max_ms": 17738.0,
+ "time_to_ready_p50_ms": 17949.7,
+ "time_to_ready_p95_ms": 26641.1,
+ "time_to_ready_max_ms": 26645.8,
+ "processing_p50_ms": 17312.3,
+ "processing_p95_ms": 17677.9,
+ "decode_ms_p50": 156.61,
+ "decode_ms_p95": 291.96,
+ "scale_ms_p50": 5794.3,
+ "scale_ms_p95": 6868.84,
+ "jpeg_encode_ms_p50": 60.28,
+ "jpeg_encode_ms_p95": 139.01,
+ "webp_encode_ms_p50": 9796.74,
+ "webp_encode_ms_p95": 11033.62,
+ "encode_ms_p50": 9844.68,
+ "encode_ms_p95": 11124.45,
+ "pipeline_ms_p50": 16593.78,
+ "pipeline_ms_p95": 17069.69,
+ "encode_share_pct": 61.4,
+ "timing_jobs": 12,
+ "original_bytes": 637620,
+ "variants_bytes": 3163980,
+ "variant_bytes": {
+ "thumbnail": 260148,
+ "crop": 78672,
+ "optimized": 2825160
+ },
+ "compression_ratio": 4.962,
"max_backlog_jobs": 12,
"max_parallel_jobs": 8,
- "rss_baseline_mib": 20.0,
- "rss_peak_mib": 1122.1,
- "rss_end_mib": 763.8,
- "rss_samples": 254,
- "rss_delta_mib": 1102.1,
+ "rss_baseline_mib": 18.5,
+ "rss_peak_mib": 949.6,
+ "rss_end_mib": 818.5,
+ "rss_samples": 264,
+ "rss_delta_mib": 931.1,
"timeline": [
{
- "t": 1.084,
- "queued": 6,
- "processing": 6,
+ "t": 1.018,
+ "queued": 7,
+ "processing": 5,
"ready": 0,
"failed": 0
},
{
- "t": 1.861,
+ "t": 1.526,
"queued": 5,
"processing": 7,
"ready": 0,
"failed": 0
},
{
- "t": 2.395,
+ "t": 2.367,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 3.195,
+ "t": 3.265,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 3.804,
+ "t": 3.777,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 4.307,
+ "t": 4.415,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 4.914,
+ "t": 4.983,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 5.49,
+ "t": 5.475,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 5.99,
+ "t": 6.15,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 6.547,
+ "t": 6.969,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 7.386,
+ "t": 7.633,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 8.066,
+ "t": 8.371,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 8.621,
+ "t": 8.987,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 9.226,
+ "t": 9.457,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 9.729,
+ "t": 10.148,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 10.222,
+ "t": 10.767,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 10.837,
+ "t": 11.329,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 11.613,
+ "t": 11.952,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 12.158,
+ "t": 12.675,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 12.822,
+ "t": 13.232,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 13.409,
+ "t": 13.711,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 13.899,
+ "t": 14.346,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 14.425,
+ "t": 14.976,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 15.051,
+ "t": 15.633,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 15.644,
+ "t": 16.204,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 16.265,
+ "t": 16.721,
"queued": 4,
"processing": 8,
"ready": 0,
"failed": 0
},
{
- "t": 17.048,
+ "t": 17.341,
"queued": 4,
- "processing": 7,
- "ready": 1,
+ "processing": 8,
+ "ready": 0,
"failed": 0
},
{
- "t": 17.714,
+ "t": 17.954,
"queued": 3,
- "processing": 5,
- "ready": 4,
+ "processing": 7,
+ "ready": 2,
"failed": 0
},
{
- "t": 18.216,
- "queued": 1,
+ "t": 18.478,
+ "queued": 0,
"processing": 6,
- "ready": 5,
+ "ready": 6,
"failed": 0
},
{
- "t": 18.891,
+ "t": 19.183,
+ "queued": 0,
+ "processing": 5,
+ "ready": 7,
+ "failed": 0
+ },
+ {
+ "t": 19.779,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 19.529,
+ "t": 20.313,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 20.017,
+ "t": 20.928,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 20.748,
+ "t": 21.683,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 21.529,
+ "t": 22.289,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 22.067,
+ "t": 22.799,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 22.652,
+ "t": 23.462,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 23.251,
+ "t": 23.964,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 23.924,
+ "t": 24.448,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 24.548,
+ "t": 25.17,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 25.071,
+ "t": 25.999,
"queued": 0,
"processing": 4,
"ready": 8,
"failed": 0
},
{
- "t": 25.888,
+ "t": 26.515,
"queued": 0,
- "processing": 3,
- "ready": 9,
+ "processing": 4,
+ "ready": 8,
"failed": 0
},
{
- "t": 26.514,
+ "t": 27.171,
+ "queued": 0,
+ "processing": 2,
+ "ready": 10,
+ "failed": 0
+ },
+ {
+ "t": 27.714,
"queued": 0,
"processing": 0,
"ready": 12,
@@ -1267,28 +2884,28 @@
{
"index": 0,
"status": 202,
- "elapsed_ms": 18.2,
+ "elapsed_ms": 26.7,
"retry_after": null,
"error_code": null
},
{
"index": 1,
"status": 202,
- "elapsed_ms": 17.4,
+ "elapsed_ms": 20.3,
"retry_after": null,
"error_code": null
},
{
"index": 2,
"status": 202,
- "elapsed_ms": 10.8,
+ "elapsed_ms": 13.8,
"retry_after": null,
"error_code": null
},
{
"index": 3,
"status": 503,
- "elapsed_ms": 3.9,
+ "elapsed_ms": 2.4,
"retry_after": "5",
"error_code": "unavailable"
}
@@ -1296,7 +2913,7 @@
"accepted_before_503": 3,
"blocked_status": 503,
"blocked_retry_after": "5",
- "recovery_seconds": 7.596,
+ "recovery_seconds": 7.653,
"recovered_jobs": 3,
"probe_status": 202,
"probe_ready": true,
diff --git a/evidence/track1.md b/evidence/track1.md
index 856c38ce..f2a0afad 100644
--- a/evidence/track1.md
+++ b/evidence/track1.md
@@ -1,16 +1,19 @@
# Lab5 track 1 — queue under load
-Generated: `2026-11-14T21:42:03.122104+00:00`
+Generated: `2026-11-20T16:02:07.856909+00:00`
Synthetic images; no tokens or credentials are stored.
-| Profile | Scenario | Jobs | Jobs/s | Queue lag p95, ms | Ready p95, ms | RSS peak, MiB |
-| --- | --- | ---: | ---: | ---: | ---: | ---: |
-| serial | small_many | 30 | 6.88 | 3194 | 3284 | 28 |
-| serial | huge_few | 12 | 0.44 | 23333 | 25521 | 205 |
-| balanced | small_many | 30 | 11.59 | 1624 | 1736 | 31 |
-| balanced | huge_few | 12 | 0.45 | 17716 | 25785 | 729 |
-| eager | small_many | 30 | 13.86 | 898 | 1158 | 33 |
-| eager | huge_few | 12 | 0.46 | 17538 | 25699 | 1122 |
+| Profile | Scenario | Jobs | Jobs/s | Queue lag p95, ms | Ready p95, ms | Encode p95, ms | Encode share, % | RSS peak, MiB |
+| --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: |
+| serial | small_many | 30 | 6.84 | 3256 | 3338 | 80 | 83.8 | 28 |
+| serial | exactly_right | 18 | 0.47 | 33591 | 35714 | 1722 | 82.0 | 174 |
+| serial | huge_few | 12 | 0.44 | 23668 | 25896 | 1508 | 63.8 | 213 |
+| balanced | small_many | 30 | 13.69 | 1147 | 1274 | 90 | 83.8 | 29 |
+| balanced | exactly_right | 18 | 0.49 | 31510 | 35764 | 6562 | 83.1 | 434 |
+| balanced | huge_few | 12 | 0.44 | 18161 | 26576 | 5606 | 61.4 | 756 |
+| eager | small_many | 30 | 15.58 | 924 | 1166 | 95 | 79.9 | 32 |
+| eager | exactly_right | 18 | 0.49 | 31735 | 35797 | 13330 | 83.1 | 818 |
+| eager | huge_few | 12 | 0.44 | 17648 | 26641 | 11124 | 61.4 | 950 |
-Backpressure: cap=3, accepted=3, first 503 Retry-After=5, recovery=7.60s, probe HTTP 202.
+Backpressure: cap=3 (deliberate), first 503 Retry-After=5, recovery=7.65s, probe HTTP 202.
diff --git a/scripts/track1_experiment.py b/scripts/track1_experiment.py
index 56731244..5c5566ab 100755
--- a/scripts/track1_experiment.py
+++ b/scripts/track1_experiment.py
@@ -1,13 +1,14 @@
#!/usr/bin/env python3
-"""Lab5 grade-5 track 1: queue behaviour under small vs huge images.
+"""Track 1: queue behaviour under small, exact and huge images.
The experiment is deliberately short and self-contained:
profiles: serial (concurrency 1, prefetch 1)
balanced (4/4)
eager (8/32)
- scenarios: small_many (30 x 192x128 PNG, variants mostly fit without scaling)
- huge_few (12 x 2400x1800 PNG, every job needs heavy scaling)
+ scenarios: small_many (30 x 192x128 PNG, almost no scaling)
+ exactly_right (18 x 1280x1280 PNG, optimized WebP at native size)
+ huge_few (12 x 2400x1800 PNG, every path needs heavy scaling)
For each cell the worker is recreated with the profile, the images are uploaded
concurrently, and the script records:
@@ -16,6 +17,8 @@ concurrently, and the script records:
* queue lag: created_at -> processing_started_at (95th percentile),
* time-to-ready: created_at -> finished_at (p50/p95),
* worker RSS baseline/peak sampled from /proc/1/status,
+ * decode / scale / JPEG / WebP timing from the worker's ready log,
+ * variant bytes versus the original (compression ratio),
* upload latency and max in-flight jobs.
A final backpressure run caps MAX_PENDING_JOBS at 3, stops the worker, shows
@@ -40,6 +43,7 @@ import json
import math
import os
import platform
+import re
import struct
import subprocess
import sys
@@ -79,7 +83,9 @@ PROFILES = [
SCENARIOS = [
{"name": "small_many", "width": 192, "height": 128,
- "description": "many small images; variants fit or crop without large scaling"},
+ "description": "many small images; almost no scaling"},
+ {"name": "exactly_right", "width": 1280, "height": 1280,
+ "description": "largest variant encodes at native 1280, so no scaling on the WebP path"},
{"name": "huge_few", "width": 2400, "height": 1800,
"description": "few 4.32 MP images; optimized variant must scale down to 1280"},
]
@@ -330,6 +336,68 @@ def wait_for_terminal(attachment_ids, started_at, timeout=240):
raise Failure(f"jobs did not finish within {timeout}s; last={rows}")
+LOGFIELD_RE = re.compile(r'(\w+)=("(?:[^"\\]|\\.)*"|\S+)')
+
+
+def job_timings(attachment_ids):
+ """Read the worker's ready log. The worker is recreated per run, so the
+ lines belong to this cell; the attachment ID is still checked explicitly."""
+ if not attachment_ids:
+ return {}
+ wanted = set(attachment_ids)
+ output = compose_capture("logs", "--no-color", "--no-log-prefix", "worker")
+ timings = {}
+ for line in output.splitlines():
+ if 'msg="attachment ready"' not in line:
+ continue
+ fields = {key: value.strip('"') for key, value in LOGFIELD_RE.findall(line)}
+ attachment_id = fields.get("attachment_id")
+ if attachment_id not in wanted:
+ continue
+ try:
+ timings[attachment_id] = {
+ name: int(fields[name])
+ for name in (
+ "decode_us", "scale_us", "jpeg_encode_us",
+ "webp_encode_us", "pipeline_us",
+ )
+ }
+ except (KeyError, ValueError):
+ continue
+ return timings
+
+
+def variant_stats(attachment_ids):
+ """Variant sizes come from the committed metadata, not from the client."""
+ if not attachment_ids:
+ return {}
+ quoted = ",".join("'" + item + "'" for item in attachment_ids)
+ sql = (
+ "SELECT a.id, a.size_bytes, "
+ "coalesce(sum((v->>'size_bytes')::bigint), 0), "
+ "coalesce(sum((v->>'size_bytes')::bigint) FILTER (WHERE v->>'name' = 'thumbnail'), 0), "
+ "coalesce(sum((v->>'size_bytes')::bigint) FILTER (WHERE v->>'name' = 'crop'), 0), "
+ "coalesce(sum((v->>'size_bytes')::bigint) FILTER (WHERE v->>'name' = 'optimized'), 0) "
+ "FROM attachments a "
+ "LEFT JOIN LATERAL jsonb_array_elements(a.variants) v ON true "
+ f"WHERE a.id IN ({quoted}) "
+ "GROUP BY a.id, a.size_bytes"
+ )
+ result = {}
+ for line in psql(sql):
+ parts = line.split("|")
+ if len(parts) != 6:
+ continue
+ result[parts[0]] = {
+ "original_bytes": int(parts[1]),
+ "variants_bytes": int(parts[2]),
+ "thumbnail_bytes": int(parts[3]),
+ "crop_bytes": int(parts[4]),
+ "optimized_bytes": int(parts[5]),
+ }
+ return result
+
+
def measure_run(base_url, actor, chat_id, profile, scenario, count, data, digest,
cpu_limit="1.0", mem_limit="2G"):
wait_drained()
@@ -371,6 +439,45 @@ def measure_run(base_url, actor, chat_id, profile, scenario, count, data, digest
if failed:
raise Failure(f"jobs failed in {profile['name']}/{scenario['name']}: {failed[:2]}")
+ row_ids = [row["attachment_id"] for row in rows]
+ timings = job_timings(row_ids)
+ if len(timings) != len(rows):
+ raise Failure(
+ f"timing logs found for {len(timings)}/{len(rows)} jobs in "
+ f"{profile['name']}/{scenario['name']}"
+ )
+ variants = variant_stats(row_ids)
+ if len(variants) != len(rows):
+ raise Failure(
+ f"variant metadata found for {len(variants)}/{len(rows)} jobs in "
+ f"{profile['name']}/{scenario['name']}"
+ )
+
+ decode_ms = [timings[item]["decode_us"] / 1000 for item in row_ids]
+ scale_ms = [timings[item]["scale_us"] / 1000 for item in row_ids]
+ jpeg_ms = [timings[item]["jpeg_encode_us"] / 1000 for item in row_ids]
+ webp_ms = [timings[item]["webp_encode_us"] / 1000 for item in row_ids]
+ encode_ms = [jpeg_ms[index] + webp_ms[index] for index in range(len(row_ids))]
+ pipeline_ms = [timings[item]["pipeline_us"] / 1000 for item in row_ids]
+ cpu_us = sum(
+ timings[item]["decode_us"] + timings[item]["scale_us"]
+ + timings[item]["jpeg_encode_us"] + timings[item]["webp_encode_us"]
+ for item in row_ids
+ )
+ encode_us = sum(
+ timings[item]["jpeg_encode_us"] + timings[item]["webp_encode_us"]
+ for item in row_ids
+ )
+ encode_share_pct = 100 * encode_us / cpu_us if cpu_us else None
+
+ variant_bytes = {
+ name: sum(item[f"{name}_bytes"] for item in variants.values())
+ for name in ("thumbnail", "crop", "optimized")
+ }
+ original_bytes = sum(item["original_bytes"] for item in variants.values())
+ variants_bytes = sum(item["variants_bytes"] for item in variants.values())
+ compression_ratio = variants_bytes / original_bytes if original_bytes else None
+
queue_lag = [
(row["processing_started_at"] - row["created_at"]) * 1000
for row in rows
@@ -446,6 +553,24 @@ def measure_run(base_url, actor, chat_id, profile, scenario, count, data, digest
"time_to_ready_max_ms": r(max(time_to_ready) if time_to_ready else None, 1),
"processing_p50_ms": r(percentile(processing, 50), 1),
"processing_p95_ms": r(percentile(processing, 95), 1),
+ "decode_ms_p50": r(percentile(decode_ms, 50), 2),
+ "decode_ms_p95": r(percentile(decode_ms, 95), 2),
+ "scale_ms_p50": r(percentile(scale_ms, 50), 2),
+ "scale_ms_p95": r(percentile(scale_ms, 95), 2),
+ "jpeg_encode_ms_p50": r(percentile(jpeg_ms, 50), 2),
+ "jpeg_encode_ms_p95": r(percentile(jpeg_ms, 95), 2),
+ "webp_encode_ms_p50": r(percentile(webp_ms, 50), 2),
+ "webp_encode_ms_p95": r(percentile(webp_ms, 95), 2),
+ "encode_ms_p50": r(percentile(encode_ms, 50), 2),
+ "encode_ms_p95": r(percentile(encode_ms, 95), 2),
+ "pipeline_ms_p50": r(percentile(pipeline_ms, 50), 2),
+ "pipeline_ms_p95": r(percentile(pipeline_ms, 95), 2),
+ "encode_share_pct": r(encode_share_pct, 1),
+ "timing_jobs": len(timings),
+ "original_bytes": original_bytes,
+ "variants_bytes": variants_bytes,
+ "variant_bytes": variant_bytes,
+ "compression_ratio": r(compression_ratio, 3),
"max_backlog_jobs": max_backlog,
"max_parallel_jobs": max_parallel,
"rss_baseline_mib": r(baseline_kb / 1024 if baseline_kb else None, 1),
@@ -637,6 +762,11 @@ def build_html(evidence):
"VmRSS процесса worker во время задания (меньше — экономнее).",
scenarios, profiles, series_for("rss_peak_mib"), " MiB", digits=0, lower_better=True,
),
+ svg_chart(
+ "Compression overhead: доля encode в pipeline CPU",
+ "encode / (decode + scale + encode), % (меньше — дешевле сжатие).",
+ scenarios, profiles, series_for("encode_share_pct"), " %", digits=1, lower_better=True,
+ ),
]
def fmt(value, digits=1, suffix=""):
@@ -653,6 +783,8 @@ def build_html(evidence):
f"{fmt(run['queue_lag_p95_ms'], 0)} "
f"{fmt(run['time_to_ready_p95_ms'], 0)} "
f"{fmt(run['processing_p95_ms'], 0)} "
+ f"{fmt(run['encode_ms_p95'], 0)} "
+ f"{fmt(run['encode_share_pct'], 1)} "
f"{fmt(run['rss_peak_mib'], 0)} "
f"{run['max_parallel_jobs']} "
f"{run['max_backlog_jobs']} "
@@ -679,6 +811,7 @@ def build_html(evidence):
f"{scenario}: пиковый RSS eager относится к serial как {ratio:.1f}×."
)
balanced = lookup.get(("balanced", "small_many"))
+ exact = lookup.get(("balanced", "exactly_right"))
huge = lookup.get(("balanced", "huge_few"))
if balanced and huge and balanced["time_to_ready_p95_ms"] and huge["time_to_ready_p95_ms"]:
observations.append(
@@ -687,6 +820,15 @@ def build_html(evidence):
"больше, чем для small_many: размер изображения меняет стоимость задания, "
"а не только число заданий."
)
+ if exact and huge and exact["encode_share_pct"] is not None and huge["encode_share_pct"] is not None:
+ observations.append(
+ "Compression vs scaling (balanced): exactly_right — encode share "
+ f"{exact['encode_share_pct']:.1f}%, scale p95 {exact['scale_ms_p95']:.0f} ms; "
+ f"huge_few — encode share {huge['encode_share_pct']:.1f}%, "
+ f"scale p95 {huge['scale_ms_p95']:.0f} ms. На exactly_right optimized "
+ "WebP кодируется в native 1280, поэтому это нижняя граница стоимости "
+ "сжатия; на huge_few к ней добавляется downscale."
+ )
huge_serial = lookup.get(("serial", "huge_few"))
huge_eager = lookup.get(("eager", "huge_few"))
if huge_serial and huge_eager and huge_serial["throughput_jobs_per_s"]:
@@ -711,10 +853,10 @@ def build_html(evidence):
"заданий в работе и в prefetch-окне брокера."
)
observations.append(
- "Backpressure: при MAX_PENDING_JOBS=3 API принял "
- f"{evidence['backpressure']['accepted_before_503']} задания, затем вернул 503 "
- f"с Retry-After={evidence['backpressure']['blocked_retry_after']}; после старта "
- f"worker очередь восстановилась за {evidence['backpressure']['recovery_seconds']:.1f} с."
+ "Backpressure: cap=3 выставлен намеренно, поэтому число принятых заданий — "
+ "следствие настройки, а не открытие. Значимо то, что API вернул 503 "
+ f"с Retry-After={evidence['backpressure']['blocked_retry_after']}, а после "
+ f"старта worker очередь восстановилась за {evidence['backpressure']['recovery_seconds']:.1f} с."
)
backpressure_rows = "".join(
@@ -786,14 +928,14 @@ code {{ background: #f1f5f9; padding: 1px 5px; border-radius: 4px; }}
{evidence['environment']['cpu_count']} CPU, Docker {html.escape(evidence['environment']['docker_version'])}.
-Гипотеза. Маленькие изображения, которые почти не требуют
-масштабирования, должны давать высокий throughput и низкий RSS на задание.
-Большие изображения требуют декодирования и downscale, поэтому каждое задание
-дороже: при том же числе worker-ов очередь растёт, p95 time-to-ready
-увеличивается, а пиковый RSS зависит от concurrency сильнее, чем у маленьких
-файлов. Одновременно проверяем компромисс: eager уменьшает ожидание,
-но повышает потребление памяти; serial экономит память, но ограничивает
-throughput.
+Гипотеза. Стоимость задания складывается из decode, масштабирования
+и сжатия. exactly_right (1280×1280) не масштабирует optimized-вариант,
+поэтому показывает нижнюю границу стоимости сжатия; huge_few добавляет
+downscale, а small_many почти не масштабируется, но и пикселей в нём
+мало. Большие изображения при том же числе worker-ов увеличивают очередь,
+p95 time-to-ready и RSS. Одновременно проверяем компромисс: eager
+уменьшает ожидание, но повышает память; serial экономит память, но
+ограничивает throughput.
@@ -813,7 +955,6 @@ throughput.
{len(runs)}
прогонов матрицы
{max(run['rss_peak_mib'] or 0 for run in runs):.0f} MiB
максимальный RSS worker
-
{evidence['backpressure']['accepted_before_503']}
принято до первого 503
{evidence['backpressure']['recovery_seconds']:.1f} s
восстановление после backpressure
@@ -823,11 +964,16 @@ throughput.
Полная матрица
-Профиль Сценарий Jobs Jobs/s Queue lag p95, ms Ready p95, ms Processing p95, ms RSS peak, MiB Max parallel Max backlog
+Профиль Сценарий Jobs Jobs/s Queue lag p95, ms Ready p95, ms Processing p95, ms Encode p95, ms Encode share, % RSS peak, MiB Max parallel Max backlog
{''.join(table_rows)}
Queue lag = processing_started_at - created_at.
Time-to-ready = finished_at - created_at. Processing = finished_at - processing_started_at.
+Encode = JPEG thumbnail + JPEG crop + WebP optimized; encode share =
+encode / (decode + scale + encode) из worker-лога.
+Суммы байт вариантов и оригинала остаются в JSON, но не выносятся в заголовок:
+синтетический PNG сжимается слишком хорошо, чтобы отношение байт было честной
+метрикой эффективности кодека.
RSS — VmRSS процесса worker, сэмплирование примерно каждые 100 мс.
Max parallel — максимум одновременно обрабатываемых заданий; max backlog —
максимум принятых, но ещё не готовых заданий.
@@ -836,8 +982,10 @@ Max parallel — максимум одновременно обрабатыва
Backpressure и восстановление
Перед этим прогоном API пересоздан с MAX_PENDING_JOBS=3, worker
-остановлен. Последовательные upload-ы показывают, где заканчивается принятая
-работа. Затем worker запускается, принятые задания должны дойти до ready.
+остановлен. Cap выбран намеренно: важно не число принятых заданий, а то, что
+API возвращает 503 с Retry-After и не теряет уже
+подтверждённую работу. Затем worker запускается, принятые задания должны дойти
+до ready.
# HTTP Latency, ms Retry-After Error code
{backpressure_rows}
@@ -857,8 +1005,15 @@ Max parallel — максимум одновременно обрабатыва
Один worker, один RabbitMQ и один PostgreSQL: это сравнение настроек, а не
HA-кластер. Числа зависят от CPU/диска конкретной машины.
-PNG-фикстуры синтетические и хорошо сжимаются; доминирует стоимость декодирования
-и масштабирования, а не передача по сети. Размер в байтах указан в JSON.
+PNG-фикстуры синтетические и хорошо сжимаются; absolute encode time зависит
+от паттерна, но сравнение сценариев при одном паттерне остаётся корректным.
+Размер в байтах указан в JSON.
+Timing-поля читаются из worker-лога. Это wall-clock внутри процесса, поэтому
+при concurrency > 1 в них входит и планировщик, что как раз и нужно для
+сравнения профилей.
+exactly_right кодирует 1.64 MP WebP, а huge_few после downscale — около
+1.23 MP. Сценарий убирает scaling, но не уравнивает объём encode; это указано,
+чтобы не сравнивать throughput как «одинаковую работу».
RSS сэмплируется из /proc/1/status; короткий прогон не заменяет
длительный memory profile и p99.
Backpressure cap приблизительный: каждая API-реплика считает pending-задания
@@ -879,20 +1034,21 @@ def write_markdown(evidence, path):
"",
"Synthetic images; no tokens or credentials are stored.",
"",
- "| Profile | Scenario | Jobs | Jobs/s | Queue lag p95, ms | Ready p95, ms | RSS peak, MiB |",
- "| --- | --- | ---: | ---: | ---: | ---: | ---: |",
+ "| Profile | Scenario | Jobs | Jobs/s | Queue lag p95, ms | Ready p95, ms | Encode p95, ms | Encode share, % | RSS peak, MiB |",
+ "| --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: |",
]
for run in evidence["runs"]:
lines.append(
f"| {run['profile']} | {run['scenario']} | {run['jobs']} | "
f"{run['throughput_jobs_per_s']:.2f} | {run['queue_lag_p95_ms']:.0f} | "
- f"{run['time_to_ready_p95_ms']:.0f} | {run['rss_peak_mib']:.0f} |"
+ f"{run['time_to_ready_p95_ms']:.0f} | {run['encode_ms_p95']:.0f} | "
+ f"{run['encode_share_pct']:.1f} | {run['rss_peak_mib']:.0f} |"
)
bp = evidence["backpressure"]
lines.extend(
[
"",
- f"Backpressure: cap={bp['cap']}, accepted={bp['accepted_before_503']}, "
+ f"Backpressure: cap={bp['cap']} (deliberate), "
f"first 503 Retry-After={bp['blocked_retry_after']}, "
f"recovery={bp['recovery_seconds']:.2f}s, probe HTTP {bp['probe_status']}.",
"",
@@ -909,6 +1065,7 @@ def main():
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--base-url", default=os.environ.get("BASE_URL", "http://localhost:8080"))
parser.add_argument("--small-jobs", type=int, default=30)
+ parser.add_argument("--exactly-jobs", type=int, default=18)
parser.add_argument("--huge-jobs", type=int, default=12)
parser.add_argument(
"--worker-cpu-limit",
@@ -941,9 +1098,14 @@ def main():
client = Client(base_url, timeout=60)
alice, _bob, _eve, chat = client.bootstrap()
+ counts = {
+ "small_many": args.small_jobs,
+ "exactly_right": args.exactly_jobs,
+ "huge_few": args.huge_jobs,
+ }
scenarios = []
for item in SCENARIOS:
- count = args.small_jobs if item["name"] == "small_many" else args.huge_jobs
+ count = counts[item["name"]]
data = make_png(item["width"], item["height"])
scenarios.append({**item, "count": count, "bytes": len(data)})
@@ -978,6 +1140,7 @@ def main():
f" throughput={result['throughput_jobs_per_s']} jobs/s "
f"queue_p95={result['queue_lag_p95_ms']} ms "
f"ready_p95={result['time_to_ready_p95_ms']} ms "
+ f"encode_share={result['encode_share_pct']}% "
f"rss_peak={result['rss_peak_mib']} MiB",
flush=True,
)
@@ -1046,9 +1209,10 @@ def main():
),
},
"hypothesis": (
- "Small images that fit the variants without large scaling should be cheaper "
- "per job; huge images force decode+downscale, so each job costs more memory "
- "and time. More concurrency should raise throughput and RSS but reduce queue lag."
+ "Job cost is decode + scale + encode. exactly_right (1280x1280) keeps the "
+ "optimized WebP at native size, so it measures the compression floor; "
+ "huge_few adds downscale, and small_many has few pixels. More concurrency "
+ "should raise throughput and RSS but reduce queue lag."
),
"profiles": PROFILES,
"scenarios": scenarios,