Backoff

Alert channel deliveries use exponential backoff so a down Slack/Discord/email provider does not get hammered.

For your jobs:

• Use a short HTTP timeout on pings (~10 seconds).
• Do not retry the ping endlessly inside the job — one attempt (or a small curl --retry) is enough; the next scheduled run will check in again.
• Set monitor grace period above p99 runtime to avoid miss storms.