Alerts for failed runs

One of the powerful things about Trigger.dev is the automatic retrying mechanics baked into your task code.

Advanced automatic retries

You can easily configure the basics like number of retries, how spaced out they are, exponential back-off etc. But you can also programmatically configure the retry behaviour based on errors. If you get an error from OpenAI that you've hit your token limit you can retry when they tell you your tokens will be refilled.

Why attempt alerts were the wrong choice

The reason I mention all of this is because you should try and make your tasks very resilient to failure using retries. If you do this, reattempts should (in most cases) mean your run succeeds in the end.

Previously we had alerts when a run attempt failed. This meant you got a lot of extra alerts that didn't mean a run had failed and it was unclear when you needed to actually pay attention.

We've changed these to be run failure alerts instead. When a run has failed and won't reattempt anymore you'll get an alert, if you subscribe.

We support three channels at the moment:

• Slack
• Email
• Webhook

You can read the docs here for how to add alerts to your project. We also support alerts for deployments succeeding and failing.