Configuration
Notifications
Alert delivery to Slack, Discord and email when insights fire. Optional — disabled when no channel is configured.
On this page
Channels
The open-source edition delivers to three channels — Slack, Discord and email — and nothing else. Each one activates when its connection details are set. You can enable one, two or all three at once.
- Slack — Block Kit messages with severity colour bar and a button that deep-links back to the resource in KubeBolt.
- Discord — Embed messages with equivalent formatting.
- Email (SMTP) — Plain-text and HTML in one message, with three delivery modes:
instant(the default),hourlydigest, ordailydigest. Supports multiple recipients.
What triggers a notification
An insight notifies when a new episode of it opens at or above the minimum
severity. A condition that is still firing, or one restored after a restart,
does not notify again, and the same insight on the same cluster notifies at
most once per cooldown. With resolved notifications on, the insight’s
resolution sends one more message, titled [Resolved].
A burst is one summary
A cluster’s first evaluation, a reconnect after its episodes expired, or a node taking dozens of workloads down can open many insights at once. Since 2.6.4 they no longer arrive as one message each:
- One summary instead. When a cluster opens 5 or more insights within 60 seconds, Slack, Discord and instant email get one summary once the cluster has been quiet for 30 seconds. A burst that does not settle gets a partial still going summary every 5 minutes at most.
- What it holds. The count by severity, every critical insight named (up to 10, each with a link), and every rule with its count and the namespaces it hit most. The first insights, sent before the burst was clear, are counted and marked.
- Below the threshold every insight is still sent the moment it is detected.
- Digests. An hourly or daily email digest with more than 20 insights is summarised the same way.
Summaries cannot be turned off. KUBEBOLT_NOTIFICATIONS_BURST_* and
KUBEBOLT_NOTIFICATIONS_DIGEST_LIST_LIMIT set the baseline, and
Administration → System → Notifications → Summaries moves it within fixed
ranges:
| Setting | Default | Range |
|---|---|---|
Burst threshold (KUBEBOLT_NOTIFICATIONS_BURST_THRESHOLD) | 5 insights | 2–1000 |
Window (_BURST_WINDOW) | 60 s | 10 s – 1 h |
Quiet before sending (_BURST_QUIET) | 30 s | 5 s – 10 min, never longer than the longest wait |
Longest wait (_BURST_MAX_WAIT) | 5 min | 30 s – 1 h |
Digest: list up to (KUBEBOLT_NOTIFICATIONS_DIGEST_LIST_LIMIT) | 20 insights | 1–500 |
An environment value past its range applies at the nearest bound. A threshold
below 2 or a list limit below 1 falls back to the default, so a 0, which once
meant off, never turns summaries off.
Global settings
Cross-channel knobs that apply to every notifier:
| Env var | Default | Purpose |
|---|---|---|
KUBEBOLT_NOTIFICATIONS_ENABLED | true | Master kill switch — set false for maintenance windows. |
KUBEBOLT_NOTIFICATIONS_MIN_SEVERITY | warning | Insights below this threshold don’t notify. Values: critical, warning, info. |
KUBEBOLT_NOTIFICATIONS_COOLDOWN | 1h | Dedup window — the same insight on the same cluster fires at most once per cooldown. |
KUBEBOLT_NOTIFICATIONS_BASE_URL | — | Public URL of your KubeBolt — embedded as a deep link in messages. |
KUBEBOLT_NOTIFICATIONS_INCLUDE_RESOLVED | false | Also notify when an insight transitions to resolved. |
KUBEBOLT_SLACK_ENABLED · KUBEBOLT_DISCORD_ENABLED · KUBEBOLT_EMAIL_ENABLED | true | Pause one channel without clearing its configuration. |
With Helm, set these through extraEnv; with Docker Compose, add them to the
API’s environment: list, which doesn’t carry them by default. Or use the
Administration page below. All variables:
Environment variables.
Always-on across clusters
Insight notifications fire for any cluster KubeBolt is evaluating, not just the one you have open:
- Agent-connected clusters (
rbac.mode=readeroroperator) are evaluated from the moment their agent connects, whether or not anyone opens them, for as long as the agent stays connected. So is the cluster KubeBolt itself runs in. - Kubeconfig contexts are evaluated while open, and keep evaluating in the background for 30 minutes after you switch away (up to eight at a time). A context nobody has opened since KubeBolt started is not evaluated.
- Metrics-only clusters have no resource views to evaluate, so they raise no insights.
The new-episode rule still applies, so restored or flapping insights don’t re-notify. See Multi-Cluster.
Slack
# Create an Incoming Webhook in Slack, then:
KUBEBOLT_SLACK_WEBHOOK_URL=https://hooks.slack.com/services/T.../B.../...
Discord
# Server Settings → Integrations → Webhooks → New Webhook, then:
KUBEBOLT_DISCORD_WEBHOOK_URL=https://discord.com/api/webhooks/...
KUBEBOLT_SMTP_HOST=smtp.example.com
KUBEBOLT_SMTP_PORT=587 # default 587 (STARTTLS); 465 = implicit TLS
KUBEBOLT_SMTP_USERNAME=alerts@example.com
KUBEBOLT_SMTP_PASSWORD=...
KUBEBOLT_SMTP_FROM=KubeBolt <alerts@example.com>
KUBEBOLT_SMTP_TO=sre@example.com,oncall@example.com
KUBEBOLT_SMTP_DIGEST_MODE=hourly # instant (default) | hourly | daily
Email counts as configured once host, sender and at least one recipient are set.
On KubeBolt Cloud, Kobi Autopilot’s recommendations don’t depend on these channels being on. When none would carry a ready recommendation, Autopilot emails it to the organization’s admins, plus a weekly digest of what is still waiting. See KubeBolt Cloud.
Admin UI: Administration → System → Notifications (/admin/system, Admin role) edits the channels and the global settings, and sends a test notification to each channel. Values saved there override the environment variables; Reset to env defaults returns to them. Requires authentication to be enabled.
Most changes apply without a restart. In the open-source edition three wait for the next one: the first channel on an install that started with none, turning the master switch back on when KubeBolt started with it off, and turning resolved notifications on when it started with them off.