Files
karaoke-gen/infrastructure/modules/cloud_run.py
T
Andrew Beveridge ☄️andClaude Opus 5.5 badd845a1d perf(jobs): move video-encoding-job (video + render) to us-east4 (v0.249.1) (#1083)
## Summary
Follow-up to #1077. `video-encoding-job` runs both the post-review
render (`render_video_worker`) and the final video (`video_worker`). In
us-central1 each run waited **2–5 min** for Cloud Run Jobs to start it,
so every customer job paid that twice. This moves it to us-east4
alongside the download, lyrics and bulk-search jobs.

## Cost (measured, not estimated)
- Over the last 30 days the job read **156 GB** from Google services,
mostly the us-central1 bucket, across 494 runs (Cloud Monitoring
`container/network/received_bytes_count`, kind=google). That's about
0.32 GB per run.
- Pricing from the Billing Catalog: GCS inter-region within North
America is **free for the first 100 GB/month, then $0.02/GB**. Cloud Run
writes to Google APIs are **$0**.
- Internet egress (224 GB/month, Dropbox etc.) and Cloud Run CPU cost
the same in either region.
- Existing cross-region reads are about 24 GB/month (GPU job and
lyrics), which the billing export shows at about $0.
- **Net: about $1.60/month**, or about $3/month worst case. Time saved:
roughly 4–8 min per customer job.

## Changes
- **backend**: `trigger_video_worker` (via `_trigger_cloud_run_job`) and
`trigger_render_video_worker` now use `cpu_jobs_region`. `GCP_REGION`
stays us-central1; encoding VM and zone logic are unchanged.
- **infra (Pulumi, already applied)**: new us-east4
`video-encoding-job`, with the legacy us-central1 copy kept for rollback
and its deletion protection turned off. A preview after applying showed
290 unchanged.
- **ci**: the video job image update now targets us-east4, using the
us-east4 CPU image copy.
- **Also updated**: rollback.sh, tests (video and render routing),
TROUBLESHOOTING/ARCHITECTURE/LESSONS-LEARNED.

## Testing
- `make test` passes: backend tests plus 1443 frontend tests.
- Prod smoke test of the new us-east4 job: it imports `video_worker` and
`render_video_worker` and calls the encoding VM's `/health` on its
public IP (port 8080 is open to 0.0.0.0/0 with API-key auth). **Started
in 25 s, exit 0.**

🤖 Generated with [Claude Code](https://claude.com/claude-code)


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

- **Updates**
- Video encoding and render jobs now run in the `us-east4` region,
alongside other CPU jobs.
- Video encoding continues to access storage and encoding resources in
`us-central1`; cross-region transfer may incur costs.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-09-29 16:33:23 -04:00

714 lines
36 KiB
Python

"""
Cloud Run resources.
Manages Cloud Run domain mapping and Cloud Run Jobs.
"""
import pulumi
import pulumi_gcp as gcp
from pulumi_gcp import cloudrun, cloudrunv2
from config import PROJECT_ID, REGION
AUDIO_WORKER_GPU_REGION = "us-east4" # L4 GPU quota available here
# Region for the latency-critical CPU jobs (audio-download, lyrics-transcription,
# bulk-search, video-encoding). Cloud Run Jobs in us-central1 queue 2-5 min before the container
# starts, regardless of image size or CPU (measured 2026-09-28 with Google's tiny
# sample job image: us-central1 4-5 min, us-east4 8-19s). Must match the
# backend's CPU_JOBS_REGION default (backend/config.py). GCP_REGION inside these
# jobs stays REGION: the bucket, queues and other resources are still there.
CPU_JOBS_REGION = "us-east4"
def cpu_job_image(region: str) -> str:
"""CPU backend image for a job in ``region``, pulled from a same-region registry.
CI pushes the CPU app image to both us-central1 (karaoke-repo) and us-east4
(the existing karaoke-backend-gpu repo, as ``karaoke-backend-cpu``) so jobs
never pull the multi-GB image across regions (Artifact Registry egress).
"""
if region == REGION:
return f"{REGION}-docker.pkg.dev/{PROJECT_ID}/karaoke-repo/karaoke-backend:latest"
if region == AUDIO_WORKER_GPU_REGION:
return f"{region}-docker.pkg.dev/{PROJECT_ID}/karaoke-backend-gpu/karaoke-backend-cpu:latest"
raise ValueError(f"No CPU backend image registry configured for region {region}")
def _job_resource_name(base: str, region: str) -> str:
"""Pulumi logical name: unchanged for REGION, suffixed for other regions so the
us-east4 job can coexist with the legacy us-central1 one during cutover."""
return base if region == REGION else f"{base}-{region}"
def create_domain_mapping() -> cloudrun.DomainMapping:
"""
Create the Cloud Run domain mapping for api.nomadkaraoke.com.
Returns:
cloudrun.DomainMapping: The domain mapping resource.
"""
domain_mapping = cloudrun.DomainMapping(
"karaoke-backend-domain",
location=REGION,
name="api.nomadkaraoke.com",
metadata=cloudrun.DomainMappingMetadataArgs(
namespace=PROJECT_ID,
),
spec=cloudrun.DomainMappingSpecArgs(
route_name="karaoke-backend",
),
)
return domain_mapping
def create_lyrics_transcription_job(
bucket: gcp.storage.Bucket,
service_account: gcp.serviceaccount.Account,
region: str = REGION,
) -> cloudrunv2.Job:
"""
Create the Cloud Run Job for lyrics transcription.
Cloud Run Jobs run to completion without HTTP request lifecycle concerns,
avoiding the instance termination issue where Cloud Run would shut down
instances mid-processing when using BackgroundTasks (e.g., job c94cc9d6).
Typical duration: 5-15 minutes (AudioShake + agentic AI correction)
Args:
bucket: The GCS bucket for job artifacts.
service_account: The service account to run the job.
Returns:
cloudrunv2.Job: The Cloud Run Job resource.
"""
lyrics_transcription_job = cloudrunv2.Job(
_job_resource_name("lyrics-transcription-job", region),
name="lyrics-transcription-job",
location=region,
deletion_protection=False, # Allow the legacy-region copy to be removed
template=cloudrunv2.JobTemplateArgs(
template=cloudrunv2.JobTemplateTemplateArgs(
containers=[
cloudrunv2.JobTemplateTemplateContainerArgs(
image=cpu_job_image(region),
args=["python", "-m", "backend.workers.lyrics_worker"],
resources=cloudrunv2.JobTemplateTemplateContainerResourcesArgs(
limits={
"cpu": "2",
"memory": "4Gi",
},
),
envs=[
# Admin token for Cloud Tasks auth header
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ADMIN_TOKENS",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/admin-tokens",
version="latest",
),
),
),
# AudioShake API token (required for transcription)
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="AUDIOSHAKE_API_TOKEN",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/audioshake-api-key",
version="latest",
),
),
),
# Cloud Run service URL for Cloud Tasks targeting
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="CLOUD_RUN_SERVICE_URL",
value="https://api.nomadkaraoke.com",
),
# Enable Cloud Tasks mode (vs direct HTTP)
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENABLE_CLOUD_TASKS",
value="true",
),
# Basic configuration
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENVIRONMENT",
value="production",
),
# Cap the duration-scaled transcription timeout below the
# task `timeout` above so the worker raises a clean
# RuntimeError before Cloud Run hard-kills the task.
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="TRANSCRIPTION_TIMEOUT_CAP_SECONDS",
value="2700",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCP_REGION",
value=REGION,
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCS_BUCKET_NAME",
value=bucket.name,
),
# Genius API token (for lyrics fetching)
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GENIUS_API_TOKEN",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/genius-api-key",
version="latest",
),
),
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GOOGLE_CLOUD_PROJECT",
value=PROJECT_ID,
),
# RapidAPI key (for Musixmatch lyrics)
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="RAPIDAPI_KEY",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/rapidapi-key",
version="latest",
),
),
),
# Spotify cookie (for Spotify lyrics)
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="SPOTIFY_COOKIE_SP_DC",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/spotify-cookie",
version="latest",
),
),
),
],
)
],
service_account=service_account.email,
timeout="3000s", # 50 minutes max per task (long mashups can take 30+ min to transcribe)
max_retries=2,
),
),
)
return lyrics_transcription_job
def create_audio_separation_job(
bucket: gcp.storage.Bucket,
service_account: gcp.serviceaccount.Account,
) -> cloudrunv2.Job:
"""
Create the Cloud Run Job for audio separation with L4 GPU.
Runs the Separator class directly with GPU acceleration, eliminating
the HTTP call to the separate audio-separator service.
Deployed in us-east4 where L4 GPU quota is available.
Typical duration: 5-10 minutes (local GPU separation)
Args:
bucket: The GCS bucket for job artifacts.
service_account: The service account to run the job.
Returns:
cloudrunv2.Job: The Cloud Run Job resource.
"""
audio_separation_job = cloudrunv2.Job(
"audio-separation-job",
name="audio-separation-job",
location=AUDIO_WORKER_GPU_REGION,
deletion_protection=False, # Allow Pulumi to replace (region change requires delete+create)
template=cloudrunv2.JobTemplateArgs(
template=cloudrunv2.JobTemplateTemplateArgs(
containers=[
cloudrunv2.JobTemplateTemplateContainerArgs(
image=f"{AUDIO_WORKER_GPU_REGION}-docker.pkg.dev/{PROJECT_ID}/karaoke-backend-gpu/karaoke-backend:latest",
args=["python", "-m", "backend.workers.audio_worker"],
resources=cloudrunv2.JobTemplateTemplateContainerResourcesArgs(
limits={
"cpu": "4",
"memory": "16Gi",
"nvidia.com/gpu": "1",
},
),
envs=[
# Admin token for Cloud Tasks auth header
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ADMIN_TOKENS",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/admin-tokens",
version="latest",
),
),
),
# Model directory (baked into GPU image, triggers local GPU mode)
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="MODEL_DIR",
value="/models",
),
# Cloud Run service URL for Cloud Tasks targeting
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="CLOUD_RUN_SERVICE_URL",
value="https://api.nomadkaraoke.com",
),
# Enable Cloud Tasks mode (vs direct HTTP)
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENABLE_CLOUD_TASKS",
value="true",
),
# Basic configuration
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENVIRONMENT",
value="production",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCP_REGION",
value=AUDIO_WORKER_GPU_REGION,
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCS_BUCKET_NAME",
value=bucket.name,
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GOOGLE_CLOUD_PROJECT",
value=PROJECT_ID,
),
],
)
],
node_selector=cloudrunv2.JobTemplateTemplateNodeSelectorArgs(
accelerator="nvidia-l4",
),
gpu_zonal_redundancy_disabled=True, # Required for GPU Jobs (capacity limitation)
service_account=service_account.email,
timeout="1800s", # 30 minutes (local GPU is faster than remote API)
max_retries=2,
),
),
)
return audio_separation_job
def create_audio_download_job(
bucket: gcp.storage.Bucket,
service_account: gcp.serviceaccount.Account,
region: str = REGION,
) -> cloudrunv2.Job:
"""
Create the Cloud Run Job for audio downloading.
Downloads audio from Spotify/YouTube/RED/OPS via the flacfetch service
(public HTTPS URL from the flacfetch-api-url secret), then triggers audio
separation and lyrics workers.
Typical duration: 30s-5 minutes (depending on source)
Args:
bucket: The GCS bucket for job artifacts.
service_account: The service account to run the job.
Returns:
cloudrunv2.Job: The Cloud Run Job resource.
"""
audio_download_job = cloudrunv2.Job(
_job_resource_name("audio-download-job", region),
name="audio-download-job",
location=region,
deletion_protection=False, # Allow the legacy-region copy to be removed
template=cloudrunv2.JobTemplateArgs(
template=cloudrunv2.JobTemplateTemplateArgs(
containers=[
cloudrunv2.JobTemplateTemplateContainerArgs(
image=cpu_job_image(region),
args=["python", "-m", "backend.workers.audio_download_worker"],
resources=cloudrunv2.JobTemplateTemplateContainerResourcesArgs(
limits={
"cpu": "1",
"memory": "1Gi",
},
),
envs=[
# Admin token for worker auth
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ADMIN_TOKENS",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/admin-tokens",
version="latest",
),
),
),
# Cloud Run service URL for triggering workers
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="CLOUD_RUN_SERVICE_URL",
value="https://api.nomadkaraoke.com",
),
# Enable Cloud Tasks mode for worker triggers
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENABLE_CLOUD_TASKS",
value="true",
),
# Basic configuration
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENVIRONMENT",
value="production",
),
# Flacfetch API key
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="FLACFETCH_API_KEY",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/flacfetch-api-key",
version="latest",
),
),
),
# Flacfetch API URL (public HTTPS endpoint; no VPC egress)
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="FLACFETCH_API_URL",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/flacfetch-api-url",
version="latest",
),
),
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCP_REGION",
value=REGION,
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCS_BUCKET_NAME",
value=bucket.name,
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GOOGLE_CLOUD_PROJECT",
value=PROJECT_ID,
),
# Email provider (Postmark). The worker's duration
# reconcile can send an over-limit cancellation email
# to the customer; without this the EmailService "fail
# loudly" guard raised in production and crashed the
# whole download (job a453d1d5).
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="POSTMARK_SERVER_TOKEN",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/postmark-server-token",
version="latest",
),
),
),
],
)
],
service_account=service_account.email,
# 20 min max. Slow torrent (RED/OPS) downloads can exceed 10 min;
# a SIGKILL at the old 600s bypassed graceful failure handling and
# left in-flight auto-retries racing manual retries (job 1cd29294).
# The worker's DOWNLOAD_WAIT_TIMEOUT_SECONDS (1080s) is kept
# strictly below this so it fails cleanly before Cloud Run kills it.
timeout="1200s",
max_retries=2,
),
),
)
return audio_download_job
def create_bulk_search_job(
bucket: gcp.storage.Bucket,
service_account: gcp.serviceaccount.Account,
region: str = REGION,
) -> cloudrunv2.Job:
"""
Create the Cloud Run Job for Bulk Mode batch audio search.
Processes every job in a batch (state_data.batch_id): runs each audio search
via flacfetch, auto-selects confident lossless matches (triggering the
audio-download-job), and parks the rest in AWAITING_AUDIO_SELECTION.
Longer timeout than the download
job because a batch can contain up to 100 songs. Idempotent across retries
(only re-processes jobs still flagged bulk_pending_search), so retries=1.
"""
bulk_search_job = cloudrunv2.Job(
_job_resource_name("bulk-search-job", region),
name="bulk-search-job",
location=region,
deletion_protection=False, # Allow the legacy-region copy to be removed
template=cloudrunv2.JobTemplateArgs(
template=cloudrunv2.JobTemplateTemplateArgs(
containers=[
cloudrunv2.JobTemplateTemplateContainerArgs(
image=cpu_job_image(region),
args=["python", "-m", "backend.workers.bulk_search_worker"],
resources=cloudrunv2.JobTemplateTemplateContainerResourcesArgs(
limits={
"cpu": "1",
"memory": "1Gi",
},
),
envs=[
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ADMIN_TOKENS",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/admin-tokens",
version="latest",
),
),
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="CLOUD_RUN_SERVICE_URL",
value="https://api.nomadkaraoke.com",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENABLE_CLOUD_TASKS",
value="true",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENVIRONMENT",
value="production",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="FLACFETCH_API_KEY",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/flacfetch-api-key",
version="latest",
),
),
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="FLACFETCH_API_URL",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/flacfetch-api-url",
version="latest",
),
),
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCP_REGION",
value=REGION,
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCS_BUCKET_NAME",
value=bucket.name,
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GOOGLE_CLOUD_PROJECT",
value=PROJECT_ID,
),
],
)
],
service_account=service_account.email,
timeout="3600s", # up to 100 songs per batch
max_retries=1,
),
),
)
return bulk_search_job
def create_video_encoding_job(
bucket: gcp.storage.Bucket,
service_account: gcp.serviceaccount.Account,
region: str = REGION,
) -> cloudrunv2.Job:
"""
Create the Cloud Run Job for video encoding.
Cloud Run Jobs run to completion (up to 1 hour), immune to Cloud Run Service
deployment rollouts. This prevents the BackgroundTask termination issue where
a deployment kills the encoding orchestrator mid-flight (see incident 2026-03-08).
The video worker orchestrates: CDG/TXT packaging, GCE encoding submission/polling,
file download from GCS, YouTube/Dropbox/GDrive upload, and Discord notification.
It needs the same env vars and secrets as the Cloud Run Service for these features.
Args:
bucket: The GCS bucket for job artifacts.
service_account: The service account to run the job.
Returns:
cloudrunv2.Job: The Cloud Run Job resource.
"""
video_encoding_job = cloudrunv2.Job(
_job_resource_name("video-encoding-job", region),
name="video-encoding-job",
location=region,
deletion_protection=False, # Allow the legacy-region copy to be removed
template=cloudrunv2.JobTemplateArgs(
template=cloudrunv2.JobTemplateTemplateArgs(
containers=[
cloudrunv2.JobTemplateTemplateContainerArgs(
image=cpu_job_image(region),
args=["python", "-m", "backend.workers.video_worker"],
resources=cloudrunv2.JobTemplateTemplateContainerResourcesArgs(
limits={
"cpu": "4",
"memory": "8Gi",
},
),
envs=[
# Admin token for worker-to-service auth
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ADMIN_TOKENS",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/admin-tokens",
version="latest",
),
),
),
# Cloud Run service URL for triggering downstream workers
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="CLOUD_RUN_SERVICE_URL",
value="https://api.nomadkaraoke.com",
),
# Distribution defaults
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="DEFAULT_BRAND_PREFIX",
value="NOMAD",
),
# Discord webhook for release notifications
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="DEFAULT_DISCORD_WEBHOOK_URL",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/discord-releases-webhook",
version="latest",
),
),
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="DEFAULT_DROPBOX_PATH",
value="/MediaUnsynced/Karaoke/Tracks-Organized",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="DEFAULT_ENABLE_YOUTUBE_UPLOAD",
value="true",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="DEFAULT_GDRIVE_FOLDER_ID",
value="1laRKAyxo0v817SstfM5XkpbWiNKNAMSX",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="DEFAULT_PRIVATE_BRAND_PREFIX",
value="NOMADNP",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="DEFAULT_PRIVATE_DROPBOX_PATH",
value="/MediaUnsynced/Karaoke/Tracks-NonPublished",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENABLE_CLOUD_TASKS",
value="true",
),
# GCE encoding worker
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENCODING_WORKER_API_KEY",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/encoding-worker-api-key",
version="latest",
),
),
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENCODING_WORKER_URL",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/encoding-worker-url",
version="latest",
),
),
),
# Multi-zone capacity-fallback VMs (JSON list).
# Without this, the encoding worker manager only
# tries the primary zone — when -c is exhausted
# the job fails. Cloud Run Service has the same
# secret mounted; the Cloud Run Job needed it too.
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENCODING_WORKER_FALLBACK_VMS",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/encoding-worker-fallback-vms",
version="latest",
),
),
),
# Core configuration
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="ENVIRONMENT",
value="production",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCP_REGION",
value=REGION,
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GCS_BUCKET_NAME",
value=bucket.name,
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="GOOGLE_CLOUD_PROJECT",
value=PROJECT_ID,
),
# Email provider for completion emails (Postmark)
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="POSTMARK_SERVER_TOKEN",
value_source=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceArgs(
secret_key_ref=cloudrunv2.JobTemplateTemplateContainerEnvValueSourceSecretKeyRefArgs(
secret=f"projects/{PROJECT_ID}/secrets/postmark-server-token",
version="latest",
),
),
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="USE_GCE_ENCODING",
value="true",
),
# The render worker also runs in this job (args
# override → backend.workers.render_video_worker)
# and triggers the video worker in-process when the
# render completes. Without these flags that
# downstream trigger would fall back to the
# deploy-vulnerable Cloud Tasks → BackgroundTask
# path this job exists to avoid.
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="USE_CLOUD_RUN_JOBS_FOR_VIDEO",
value="true",
),
cloudrunv2.JobTemplateTemplateContainerEnvArgs(
name="USE_CLOUD_RUN_JOBS_FOR_RENDER",
value="true",
),
],
)
],
service_account=service_account.email,
timeout="3600s", # 1 hour max per task
max_retries=2,
),
),
)
return video_encoding_job