Heroku Scheduler Monitor vs. One-off Dyno Metrics: Which Add-On to Use

You run one-off dynos on your Heroku app, and you want to know when one fails, runs too long, or stops before it's done.

We build two add-ons for that: One-off Dyno Metrics and Heroku Scheduler Monitor. Both watch detached one-off dynos and send you alert notifications.

Heroku Scheduler Monitor vs. One-off Dyno Metrics comes down to the kind of one-off dynos you run. If they're all Heroku Scheduler jobs, choose Heroku Scheduler Monitor. If you also want to monitor other kinds, such as release phase, heroku run:detached or Platform API dynos, choose One-off Dyno Metrics. The rest of this post covers the differences in detail.

Which one-off dynos each add-on monitors

One-off Dyno Metrics monitors every detached one-off dyno, whatever started it, while Heroku Scheduler Monitor monitors only scheduler dynos and ignores the rest. Detached one-off dynos run in the background and write their output to your app logs. The first part of a dyno name is its type, such as scheduler in scheduler.1234, and the type tells you what started it:

Started byDyno typeOne-off Dyno MetricsHeroku Scheduler Monitor
Heroku SchedulerschedulerYesYes
heroku run:detachedrunYesNo
Release phasereleaseYesNo
Our Advanced Scheduler add-onadvanced-schedulerYesNo
The Platform APIThe type you setYesNo

Heroku Scheduler Monitor "only considers metrics from detached scheduler one-off dynos started by Heroku Scheduler". One-off Dyno Metrics measures every source in the table, including release phase dynos, and each monitor's Dyno Type filter picks the ones you care about.

What Heroku Scheduler Monitor adds for Scheduler jobs

Because it watches only Heroku Scheduler jobs, Heroku Scheduler Monitor adds three things: skip detection, no filters to set, and a simpler setup.

Skip detection: Heroku Scheduler job execution "is expected but not guaranteed". With skip detection, Heroku Scheduler Monitor notices a run that never started, including one that couldn't start because the app was at its one-off dyno limit, and sends you an alert notification. One-off Dyno Metrics works from the one-off dynos that start, so a run that never started gives it nothing to measure.

No filters to set: every plot and monitor in Heroku Scheduler Monitor counts only scheduler dynos. In One-off Dyno Metrics you get the same focus by setting each monitor's Dyno Type filter to "scheduler", in double quotes. Without the quotes, the filter also matches advanced-scheduler.

Simplicity over granularity: because One-off Dyno Metrics takes every type of one-off dyno into account, it allows more granular monitoring and alerts. Each monitor can filter on dyno type, command and size, and you can create several monitors of the same type side by side for a detailed setup. Heroku Scheduler Monitor is more streamlined: it trades that flexibility for simplicity, both when you set it up and in daily use.

What the two add-ons share

Beyond those differences, both add-ons plot Events, Execution Time, Throughput by exit status, and Concurrency. By default, alerts go to the app owners and collaborators, or to the team admins for an app in a Heroku team, and you can add more addresses or an alias. Both are fully configured as soon as you add them and pick up their data from your app automatically.

Next steps

You now know which add-on fits your one-off dynos, so the next step is to try it on your own app. If all of them come from Heroku Scheduler, install Heroku Scheduler Monitor. If you run other kinds too, install One-off Dyno Metrics. Either one starts collecting data as soon as you add it, so your next runs show up in its dashboard without any setup.

If you want to keep reading, Heroku Scheduler best practices for production covers how to write jobs that are safe to run twice, catch up after a missed run, and shut down cleanly when Heroku stops the dyno.