Incident emails for disconnected inboxes
When an inbox stops working, for example a Telegram bot whose token was revoked, the workspace's owner gets an email that says what broke and how to fix it, with reminders that back off instead of repeating, an Inspect button that opens the inbox and an Acknowledge button that stops the reminders.
Last updated
When an incident is raised
A disconnected inbox is an incident of its workspace. One is opened when:
- a channel refuses the inbox's credentials (a Telegram bot token revoked in BotFather, a mailbox sign-in that ended), found by the five-minute health check, or the inbox's installation needs a reconnect;
- an inbox has been in error for 30 minutes for any other reason; the provider's message goes in the email.
Each outage of each inbox is its own incident, with its own schedule and acknowledgement. It clears when the inbox is out of error: fixed, paused, archived or deleted. The banner on the inbox's page and the in-app notification to everyone who manages inboxes stay as before; the email is added on top.
Who is emailed
- The workspace's owner, while they are an active member. A workspace whose owner has left emails its active members who manage inboxes (
inboxes:manage) instead. - Only people of that workspace, and only about that workspace's inboxes: never another workspace's people.
- Each recipient can turn their own copy off under Preferences › Notifications (A disconnected inbox (alert emails), email on by default). These emails go at once and are never held for the hourly digest.
How often
Instead of an email every few minutes for as long as the problem lasts, the reminders back off. With the default schedule an incident emails:
- at once;
- after 1, 4 and 12 hours;
- a last time after 24 hours, saying no more emails will come until it is fixed;
- once more, Resolved, with how long it lasted, when the inbox works again (only if at least one email went out).
Several inboxes that break together are one email listing each of them (its channel, since when and what the provider said), and the resolved ones are listed together too.
A workspace administrator (settings:manage) chooses under Settings › Notifications › Alert emails whether these emails go at all and on which schedule: Default (1h, 4h, 12h, 24h), Fewer (4h, 24h) or Only once. Through the API, a schedule can have at most eight emails, at least an hour apart, the last within 7 days.
Inspect and Acknowledge
- Inspect opens the inbox's page (the inbox list when the email names several), where the banner says what to do for that channel; for Telegram, paste the new token from @BotFather into the Bot token card.
- Acknowledge says someone is on it: no more reminders for that outage, for anyone. The Resolved email still comes. The inbox page and the inbox list then say who acknowledged it and when, and the inbox page's banner has the same button.
Not covered yet
Other kinds of incidents (a failing webhook endpoint, a paused bot) keep their existing one-time notices. There is no per-workspace mute yet, and these emails are in English, like Kommio's other system emails.
Self-hosted operators: platform alerts in the platform console follow the same kind of schedule; see the self-hosted settings.
Every claim on these pages is covered by a test in the repository, and the API reference is generated from the same contract every server serves at /api/openapi. Corrections to [email protected].