Run Notifications
Run Notifications tell you when something you started with Run Now has finished, without you having to sit on the Scheduler page waiting for it. When the run ends, a notification appears in the corner of the screen saying what finished and whether it worked.
They cover both kinds of Run Now:
A Job, started from Scheduler > Jobs or the Scheduler page.
A config - a Load, Stage, Archive, Code, Export or Stored Procedure - started from its own page or list.
Only the person who started the run sees the notification. It follows you around the app, so you can start a run, carry on working somewhere else in Bragi, and still be told when it is done.
Turning them on
Run Notifications are off until you turn them on. Switch on Run Now Notifications on the Settings page and save.
Each Environment has its own switch, because each has its own Scheduler and they are not all set up the same way. Turning it on for one Environment says that that Environment's Scheduler can reach Bragi; it says nothing about the others.
There is no URL to configure with it. Bragi registers the notification while it is serving you, so it already knows the address you reached it on, including its port and any tenant path, and points the Scheduler back at that.
Runs themselves are unaffected either way.
Use Test Notification below the switch to check the path before relying on it. The switch has to be on and saved first, since the test takes the same route a real notification does.
When the Scheduler needs a different address
The Scheduler is what posts the notification, and it often runs on a different machine from the app. Usually it reaches Bragi on the same address you do. When it cannot - the app sits behind a reverse proxy the Scheduler cannot use, or the Scheduler is on a network where that name does not resolve - set Run Notification URL Override for that Environment on the Settings page, for example https://bragi.internal. Leave it blank otherwise.
Each Environment has its own override, so an Environment whose Scheduler sits in a different network can point somewhere else without affecting the others.
Testing it
On the Settings page is a Test Notification button. It asks the Scheduler for that Environment to post a test notification to Bragi, using the same address a real notification would use. The Scheduler does the posting rather than the app, so the test proves the path a real notification would take rather than only that Bragi can reach itself.
The Scheduler picks the test up on its next tick, so allow a few seconds for an answer. What comes back tells you which part of the path is at fault:
Result | What it means |
|---|---|
The Scheduler reached Bragi and the notification was accepted | Everything works. A notification also appears on screen, which is the same notification a finished run would raise. |
The Scheduler picked the test up but could not deliver it | A Scheduler is running, but it cannot reach Bragi on the address it was given. The error it got is shown. Check that any firewall or proxy between the Scheduler and the app allows it, that the certificate is trusted on the Scheduler host if the address is |
No Scheduler picked the test up | No Scheduler is running for this Environment, or it is not working. Check Scheduler Health. |
A test that is not picked up within ten minutes is abandoned rather than answered later, so a Scheduler that comes back up hours afterwards does not post a notification for a test you have long since forgotten about.
What you see
The notification names what ran and whether it worked. A failed run also names the tasks that failed, so you know where to look on the Scheduler page.
Notifications are transient. They are shown to browser sessions you have open at the time, so a run that finishes after you have closed Bragi is not waiting for you when you next sign in. The run itself, and its outcome, are recorded as usual and can be seen on the Scheduler and History pages.
A run always goes ahead whether or not it can be announced. If the URL is wrong, or the app cannot be reached when the run ends, the work still runs and the failure to notify is logged rather than raised to the user.