Cloud Console Own Projects

User manual and SOP

The runbook for Cloud Console: how to get in, how to find your way around, and how to do the daily work step by step. Each task is a list of numbered steps beside a picture of the real screen, with the same numbers on the controls to use, and then what you see when it worked. The last part lists the messages the console shows, what each one means and how to fix it.

How to read the steps

The pictures are of a local copy of the console with a made-up project, shop, and a made-up admin. Your console has your own projects and names. Choose Dark in the top bar of this page to see the pictures in the dark theme.

The parts of the manual

Getting started and signing inYour console's address, an invitation, Sign in with Google, the break-glass admin and its authenticator, passkeys, signing out. Finding your wayThe top bar, the sidebar groups, breadcrumbs, toolbars, tabs, row menus and notices, the phone menu, and why what you see depends on your permissions. Cloud TasksCreate a queue, add a task or up to 50 from a file, find a task and its attempts, retry or delete failed tasks, pause, resume or purge a queue. Cloud SchedulerCreate a job in three steps, force run, pause and resume it, and set up an uptime check. Project settingsAPI keys, secrets and their allowed origins, service accounts, alerts and quiet hours, and export and import of a project's settings. Day to dayRead the dashboard, check usage against the daily quota, switch maintenance on for a deploy, and turn on the daily summary email. For adminsInvite a person and set their permissions, read the audit log, and set a project's daily cap. Troubleshooting and FAQsEvery common message, what it means and how to fix it; why a button is missing, why a task has not run, how long things are kept.

Standard procedures

The routines most people repeat. Each line links to the walkthrough that shows it step by step.

The morning check

Any member, a few minutes.

  1. Open the Dashboard. Read the Last 24 hours tile: No failures in the last 24 hours is what you want.
  2. Check that the Cron tile says Running.
  3. Look at Failing jobs and Backlogs. Open each row that is there.
  4. Open Usage. Every line should say OK.
  5. For a failure, follow A task failed for good or A job failed.

Before and after a deploy

Someone with the pause permission in the project.

  1. On Projects, click Hold dispatch for the project, and confirm (how).
  2. Check the banner Maintenance is on for …: its queues and jobs are held at the top of the page.
  3. Run the deploy or the migration. New tasks are still accepted and wait.
  4. Click Resume dispatch. The banner goes.
  5. Check on the Dashboard that Backlogs empties.

A task failed for good

Someone with the run or delete permission.

  1. Open the queue's Failed tab (how).
  2. Open the task and read its Attempts: the code, the error and the answer (how).
  3. Fix the cause: the URL, the secret or its allowed origins, or the app that answers.
  4. Click Retry all, or Retry in the task's row menu. Or delete what should not run again.

A job failed

Someone with the run permission.

  1. Open the job from Failing jobs on the Dashboard, or from Jobs.
  2. On its Logs tab, click Show attempts on the failed run and read the error.
  3. Fix the cause, then click Force run (how).
  4. A run that says Skipped was not a failure: the label says why (see the FAQ).

A key or secret is expiring

Someone with the keys or secrets permission.

  1. The project's alert channels get a warning 7 days and 1 day before the end.
  2. Click Rotate on the key or secret (keys, secrets).
  3. Put the new key in the app, or move the jobs, tasks and alerts that name the old secret to the new name.
  4. Revoke the old key, or delete the old secret, once nothing uses it.

A person joins or leaves

Admins.

  1. To add someone, invite them with only the permissions they need. They accept within 7 days.
  2. To change what someone may do, use Edit access in their row on People.
  3. When someone leaves, choose Disable or Remove in their row. Either signs them out everywhere.
  4. Check the audit log for the change.

Words used here

WordWhat it is
ProjectOne app's own space, such as shop: its queues, jobs, API keys, secrets and alerts. You see only the projects you have a permission on.
QueueA named line of HTTP tasks, with a rate limit and retry settings. Every project has a queue named default.
TaskOne HTTP call the queue makes: a method, a URL, headers, a body and an auth option. It retries as its queue says.
JobA call on a schedule (unix-cron, in a time zone): to a URL, to a queue (it adds a task), or an uptime check.
AttemptOne try of a task or a job run: its time, its answer's status or error, and how long it took.
Daily quotaHow many task creations, job runs and task retries a project may make in one UTC day. 5,000 unless an admin sets another cap.
MaintenanceA switch that holds a project's sending (or the whole console's) during a deploy. Tasks are still accepted and wait.

Cloud Console user manual, from the console as it is on main. The pictures are of a local copy with a made-up project, shop, and a made-up admin. 22 Oct 2026.