Operations
Anything that builds, installs, syncs or removes runs as an operation. You press the button, the platform does the work, and the operation tells you how it went.
On this page
Why operations
Building a cluster takes minutes, not milliseconds. So the cabinet does not make you watch a spinner: it starts an operation and lets you go. Close the tab, have lunch — the work keeps going and Activity keeps the record.
Following one
Every operation has an id (run-…) and a page with its steps: Preparing the machines, Applying the base applications and so on. The current step says what it is doing in a sentence. A cluster's row shows a progress bar with the same phases.
| Status | Meaning |
|---|---|
| queued | Waiting for a free slot — it starts by itself. See Limits. |
| running | Working on it. |
| succeeded | Done. |
| succeeded with warnings | Finished, but part of it did not go through. The steps say which. |
| failed | Stopped. The page says why and what to do. |
| cancelled | Somebody stopped it. What it had already done stays. |
When one fails
The failed operation's page says three things: what kind of failure it was, the step it failed at, and what to do about it. Every failure has an error code, and every code has a page here under Troubleshooting — the cabinet links to it with What to do →.
Often the fix is on your side (a revoked token, a machine type out of stock); sometimes it is ours. The page says which, so you never wonder whether to wait or to act.