Skip to content

Schedules (cron) ​

Settings → Schedules controls the two automatic jobs. Both can be enabled/disabled individually and their timing configured; a global switch stops everything if needed.

JobDefaultTask
Snapshotdaily at 23:00collects all portals and overwrites today's preview with the final daily state (details)
Invoice runmonthly, on the 1stcreates the PRE-invoices for the previous month — the month before the 1st in the schedule's timezone — and replaces its own not-yet-finalized drafts (details)

A job is due from its time on for the rest of that day, once: if Multiverse was not running at the scheduled minute (a restart, a deployment) or a manual run of the same kind was still in progress, the job starts as soon as it can — the invoice run on the 1st only. A run that started and never finished (Multiverse was restarted or redeployed in the middle) is started again, three times at most; after that the audit log says it was given up. The same goes for a run that could not do its work because the database could not be reached at that moment: it is made up for as soon as the database is back. A run that went through but ended with errors for single customers is done — its audit log entry is a failed one and names them; run those customers by hand. An invoice run interrupted on the 1st is taken up again — the same day or on the following days of the month — and carries on where it stopped: it keeps every invoice that was created since the run first started (by the run itself, or by hand), replaces the older drafts of the customers it had not reached yet, and creates what is still missing. If Multiverse was stopped while one customer was being written to weclapp, the next run names that customer: weclapp may hold a draft for it that the invoice list here does not show — it would be a second one. While a deployment is rolling out, two instances of Multiverse run side by side for a moment; a job that finds its run still working in the other instance waits for it instead of starting beside it. A nightly snapshot that did not run to its end before the day was over is not made up for — the audit log says so. Saving a schedule whose time has already passed today does not start the job at once (it runs at its next regular time), and it does not cancel a run that is still owed. The audit log entry of the invoice run names every portal it had to bill from an old snapshot (details).

Email when a scheduled run goes wrong ​

A run that went wrong is a failed entry in the audit log — which nobody reads every morning. In the settings, below the schedule of the nightly snapshot (Settings → Snapshots), the field Recipients takes one or several email addresses, separated by comma or semicolon (ten at most). Multiverse then sends an email through the SMTP connection whenever one of the two automatic jobs did not go through cleanly. An empty field sends nothing.

JobAn email is sent when …It names
Invoice runthe run ends with errors for single customers or with warnings; an invoice was created with a note (a figure of an account the snapshot could not read, a position priced 0,00 €); positions are billed with the month's peak — more than the month ended with; it failed before it could finish (one email per attempt); it was given up after three attempts; another instance of Multiverse took the run overthe month, how many invoices were created, every customer with an error and its message, every warning, every invoice created with a note, and — in a list of their own — the positions billed with the month's peak
Snapshota portal could not be read; the night's snapshot was given up after three attempts, or did not finish before the day was overthe portals and their error messages

The positions billed with the month's peak are no error: a month bills the highest figure of any of its days (quantity determination), so in most months some positions are billed with more than the month ended with, and the email of the 1st lists them — the subject then reads "… — 3 position(s) billed with the month's peak" and the text says that nothing went wrong. The list is for a look before the invoices are booked: a figure that was wrong for one night only (a customer merged by mistake, a value corrected the day after) looks the same.

No email is sent for a run that went through cleanly and bills no position above its month-end figure, for a manual snapshot or invoice run (its result is on the screen), for what a snapshot notes about its own figures (the invoice run that reads that day repeats those notes as warnings), and for an installation without a weclapp connection.

The email about an invoice run ends with what to do about it: a manual invoice run for that month without "Overwrite existing" creates only what is missing and leaves every existing invoice of the month as it is.

An email never changes a run. It is sent after the run has written its audit log entry, and nothing waits for it: if it cannot be sent — no SMTP connection is stored, the mail server refuses, a timeout — the audit log gets a failed entry of its own (action cron-mail), and the run counts exactly as it would without the field. While the database cannot be reached, no email can be sent either (the recipients and the SMTP connection are stored there). The email is a hint; the audit log stays the record.

Manual still possible

Disabled jobs only remove the automation — manual snapshots and manual invoice runs keep working. If the snapshot job is disabled, the manual snapshot is the authoritative daily state.