Backup jobs

Backup health gathers jobs from every integration that offers them. Return backups beside widgets. It answers with a list of jobs; Valet works out what’s late, failing or oddly sized.

async backups() {
  const jobs = await api('/jobs');
  return jobs.map(j => ({
    name: j.name,                       // the only thing required
    lastSuccess: j.lastGoodAt,          // ms or an ISO date
    lastRun: j.lastRunAt,
    result: j.ok ? 'ok' : 'failed',     // or 'warning', 'running'
    error: j.error,
    size: j.bytes,                      // the whole of what it keeps, not what one run added
    next: j.nextAt,
    every: j.intervalMs,
  }));
}

Field
name	the job, plan or source
source	what runs it, e.g. “Kopia”; your integration’s name if left out
lastSuccess	when it last finished without an error
lastRun, result	when it last ran at all, and how: ok, warning (worked, worth a look), failed (a problem until a run works) or running
error	why it failed, in the service’s words
size, sizes	bytes now, and earlier sizes oldest first if the service keeps them. Otherwise Valet remembers
next	when it runs next
every	ms it should never go without a success, if the service knows its schedule. Without it, the gap from lastRun to next is used, then backups.overdue
href	where to look at it

Leave out what the service doesn’t say. A job that doesn’t fit the shape is left out, never trusted, and a backups() that throws is shown as “couldn’t ask”.

In a built-in service(), backups: (d, o) => [...] reads the same answers as the card, and backupGet adds calls made only when backups are asked. The sample must give at least one job: npm test checks it.