Cloud tasks queue setup
Skill jeremylongshore/claude-code-plugins-plus-skills/skills/14-gcp-skills/cloud-tasks-queue-setup
'Configure cloud tasks queue setup operations. Auto-activating skill for GCP Skills.From its SKILL.md
npx -y skills add jeremylongshore/claude-code-plugins-plus-skills --skill cloud-tasks-queue-setupAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its file declares
Copied from the file, not written here
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
3.9 KB, 742 tokens by cl100k_base, as published. Nobody here has run it
Cloud Tasks Queue Setup
Overview
Create and configure Google Cloud Tasks queues for reliable asynchronous task execution. Covers queue creation with rate limiting and retry policies, HTTP and App Engine task targets, task dispatching with scheduling delays, and queue management operations (pause, resume, purge).
Prerequisites
- Google Cloud project with Cloud Tasks API enabled (
gcloud services enable cloudtasks.googleapis.com) gcloudCLI authenticated withroles/cloudtasks.adminorroles/cloudtasks.enqueuerIAM role- Target HTTP endpoint or App Engine service to receive dispatched tasks
- Service account with appropriate permissions for task execution
Instructions
- Create a Cloud Tasks queue with rate limiting:
gcloud tasks queues create <queue-name> --location=<region> --max-dispatches-per-second=10 --max-concurrent-dispatches=5 - Configure retry policy for failed tasks:
gcloud tasks queues update <queue-name> --max-attempts=5 --min-backoff=1s --max-backoff=300s --max-doublings=4 - Create an HTTP task targeting your endpoint:
gcloud tasks create-http-task --queue=<queue-name> --url=https://your-service.run.app/process --method=POST --body-content='<json-payload>' - Schedule a delayed task by adding
--schedule-timewith an ISO 8601 timestamp up to 30 days in the future - Verify queue status and task counts:
gcloud tasks queues describe <queue-name>to check dispatch rate, retry config, and queue state - Manage queue operations: pause (
gcloud tasks queues pause), resume (gcloud tasks queues resume), or purge all tasks (gcloud tasks queues purge)
Examples
Setting up an email processing queue: Create a queue with 5 dispatches per second rate limit and 3 max concurrent tasks to avoid overwhelming the email service. Configure retry with exponential backoff (1s min, 60s max, 3 doublings) and 10 max attempts for transient failures.
Scheduling delayed webhook callbacks: Create HTTP tasks with a 15-minute schedule delay to implement webhook retries. Each task POSTs to the callback URL with the original event payload and includes an OIDC token for authentication.
Output
- Cloud Tasks queue created with rate limiting and retry configuration
- Task creation commands for HTTP and App Engine targets
- Queue management commands for operational control
Error Handling
| Error | Cause | Solution |
|---|---|---|
| Queue already exists | Queue name taken in the specified region | Use gcloud tasks queues describe to check existing config; update if needed |
| PERMISSION_DENIED on task creation | Service account lacks cloudtasks.tasks.create permission | Grant roles/cloudtasks.enqueuer to the service account |
| Task handler returns 5xx | Target endpoint failed to process the task | Tasks auto-retry per queue retry policy; check handler logs for root cause |
| Rate limit exceeded | Dispatch rate exceeds queue configuration | Increase max-dispatches-per-second or reduce task creation rate |
Resources
- Cloud Tasks documentation: https://cloud.google.com/tasks/docs
- Queue configuration reference: https://cloud.google.com/tasks/docs/configuring-queues
- Creating HTTP tasks: https://cloud.google.com/tasks/docs/creating-http-target-tasks
- IAM permissions: https://cloud.google.com/tasks/docs/access-control
Related Skills
Part of the GCP Skills skill category. Tags: gcp, bigquery, vertex-ai, cloud-run, firebase
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.