# Alerts you can rely on, and no lost exceptions

Published on 21 November 2026 at https://www.larabug.com/changelog/alerts-you-can-rely-on-and-no-lost-exceptions

The first week after the launch was mostly about alerts: which ones arrive, how many of them, and channels that stay on. Exceptions that did not fit a column are no longer lost on the way in, and the application page is faster on a busy project.

### Alerts

- **Milestones only works again.** The setting was ignored, so every application got milestone mail whatever its box said. It is read again: ticked mails the 1st, 10th, 100th occurrence and so on, unticked mails every one. Every existing application is ticked, which is what it was getting, and new applications start ticked.

- **A burst arrives as one mail.** When something other than an exception opens a run of issues at once, such as every scheduled command failing while a database is away, an email channel now gets one mail listing them a minute later instead of one mail each. Exceptions are still mailed one by one, as they arrive.

- **One mail per address.** An address reached by two email channels got every mail twice. It now gets it once.

- **Channels time out.** Slack, Discord, Teams, webhook and push deliveries give up after 10 seconds instead of waiting on an endpoint that never answers, and your own SMTP server gets 10 seconds too. A hang now counts as a failed delivery.

- **Rate limits no longer switch a channel off.** A 429 from Discord, Slack, Teams or a webhook waits for the time it names and never counts as a failure. A delivery that gets through lifts an automatic switch off, the owner gets a mail when a channel is switched off, and the settings show why, with Send a test and turn on.

### Exceptions and usage

- **No more lost exceptions.** An occurrence whose user, host or message did not fit its column, or whose message held an emoji, could be dropped on the way in. Every value is now fitted to its column, and what was cut is marked (truncated) on the occurrence.

- **Failed jobs past the telemetry allowance.** Once a month's telemetry is used up, a failed job's exception still arrives as an issue. The job itself is not stored.

- **The 100% mark** in the usage mail and banner now means ingest is refusing, not 99.5%.

- **Queue job times** from servers ahead of UTC are filed at the moment they ran, not hours later.

- **Renamed scheduled tasks.** A task that was renamed or removed is no longer reported as stopped while the scheduler keeps running.

### The panel

- **Failing means an unread issue.** A failed job alone no longer marks an application Failing for a day. The day's failed jobs are listed in the status detail.

- **A faster application page.** The queue figures and the newest occurrences no longer read every job and occurrence, so the page loads far faster on a busy application, and its Queued and Failed jobs figures no longer read 0.

- **Long exception messages** are held to four lines with Show full message and a copy button. A query exception shows the database's own words first and the SQL behind the toggle. Slack, Teams and Discord alerts cut the message at 1,000 characters.

- **Billing details live on your profile.** Subscribing with incomplete details takes you there and back. A VAT number is checked with VIES once per save, and a slow VIES no longer holds up the form.

## What changed

- Milestones only works again, and is on for every application (Fixed, Panel)
- Rate limits from Discord, Slack, Teams or a webhook no longer switch a channel off (Fixed, Panel)
- An address reached by two email channels gets each mail once (Fixed, Panel)
- Exceptions with an oversized field or an emoji are no longer lost on ingest (Fixed, Panel)
- A failed job's exception still arrives after the telemetry allowance is used up (Fixed, Panel)
- The 100% usage mail and banner wait until ingest actually refuses (Fixed, Panel)
- Queue job times from servers ahead of UTC are filed at the right moment (Fixed, Panel)
- A renamed or removed scheduled task is no longer reported as stopped (Fixed, Panel)
- Queued and Failed jobs on the application page no longer read 0 on a busy application (Fixed, Panel)
- Google's round favicon crop shows the whole bug (Fixed, Site)
- A burst of issues that are not exceptions arrives on an email channel as one mail (Improved, Panel)
- Notification channels time out after 10 seconds instead of hanging (Improved, Panel)
- Failing means an unread issue, not a failed job in the last day (Improved, Panel)
- The application page loads far faster on a busy application (Improved, Panel)
- A long exception message is held to four lines, with Show full message and a copy button (Improved, Panel)
- Billing details on your profile, and one VIES check per save (Improved, Panel)
- Turn on and Turn off on every channel, the reason it was switched off, and a mail to the owner when it is (Added, Panel)
- Fields cut to fit are marked (truncated) on an occurrence (Added, Panel)
