Operations

Anything that takes longer than a request is recorded as an operation: moving a machine between hosts, capturing and restoring a snapshot, publishing or importing an image, building a pool, exporting a backup. Operations under System is where they all appear, so you do not have to find each one on the page of the thing you changed.

An operation is a durable record, not a progress bar. It survives a reload of the console and a restart of Virtainer Free, which is the only reason a change that takes twenty minutes can be left running while you do something else.

Reading the list

Each row carries the operation’s type, what it was done to, its current state, and the step it is on. The chips above the list filter by state (All, In progress, Uncertain, Failed, Succeeded), and the dropdown beside them filters by area (Instances, Images, Storage, Networks). Live keeps the list updating while it is on.

Open a row for the full record: the type, the target, the step, and when it was created, started and finished. While something is running the timing reads Running for; once it ends it reads Took. The raw message the host produced is on the detail page too, which is the text to paste when you ask for help.

The states

State	What it means
In progress	The host is working on it, and the step it names is where it is
Succeeded	The host completed it and confirmed the result
Failed	It stopped, and the record carries the reason. A failure is an answer, not a loose end
Cancelled	It was stopped before finishing
Uncertain	The host cannot say which of the above is true

Uncertain is not a failure

An operation becomes Uncertain when the host lost the thread while it was running, most often because Virtainer Free restarted, or the host restarted, partway through. The work may have finished, may have partly finished, or may never have started. The host refuses to guess, and says so, rather than reporting a result it did not observe.

This is the state to act on. Where the host can ask the underlying system again, the row offers Recheck, and Recheck all does every uncertain operation at once. A recheck resolves the record to a real state; it does not start the work over.

Where a recheck cannot settle it, the row says what to do next instead of leaving you with a warning colour. Read that line before retrying the original action, because retrying something that partly completed is not always safe.

Cancelling

An operation that is still running can be cancelled. Cancellation means the host stops driving it forward; it does not promise that a partially applied change rolls itself back. For anything that writes to disks, check the state of the machine or pool afterwards rather than assuming the world is as it was before you started.

Why this page exists

The alternative is a console that reports progress in a dialog on the page you happened to be on, and loses the answer when you navigate away. Because every long action lands here instead, you can start something, close the tab, and come back to find out what happened, including after a host reboot that would otherwise have destroyed the only record of the attempt.

Snapshots, machine transfer, backup and image publishing each have their own page describing what they do. This page is where you find out whether the one you started finished.