Maintenance Windows
Deploy without
waking the on-call.
Suppress alerts during planned maintenance, deployments, and reboots. Set windows per-node or workspace-wide. Scheduled or on-demand. When the window ends, full detection resumes automatically, with no manual re-enable required.
How It Works
Suppress alerts without stopping detection
Maintenance windows suppress alert delivery and escalation. Detection continues to run and incidents are still recorded in the audit log, so you have a complete history of what happened during maintenance, but your team isn't paged for expected traffic changes.
Scheduled windows
Set a recurring maintenance window (e.g., every Sunday 02:00–04:00 UTC) for regular deployments or traffic pattern changes. Fires automatically every week with no manual action.
On-demand windows
Start an immediate maintenance window from the Console, CLI, or API for unplanned work. Set a duration (15 minutes to 24 hours) or stop it manually when done.
Per-node scope
Apply a window to a single node, like your database while you run a migration, without affecting detection on the rest of your fleet.
Workspace-wide scope
Apply a window to all nodes simultaneously for infrastructure-level events: provider maintenance, major releases, or planned traffic experiments.
Console & API
Set from Console, CI/CD, or CLI
Maintenance windows can be set from the Flowtriq dashboard with a date picker or from the CLI with ftagent --maintenance. For CI/CD pipelines, use the CLI in your deploy scripts: start a window before the deploy runs, end it when the deploy completes.
All window operations are logged in the immutable audit trail: who created it, when it started, when it ended, and whether it was ended manually or automatically.
Start a maintenance window via CLI
ftagent --maintenance start \ --node fra-core-01 \ --duration 30 \ --reason "Deploy v2.4.1"
Node: fra-core-01
Duration: 30 minutes
Reason: Deploy v2.4.1
By: deploy-bot
14:04:19 PPS=38,201 ELEVATED
→ Incident opened · b9e1f4a2
Alert suppressed · in maintenance
Logged to audit trail
14:30:00 Maintenance window ended
✓ Full detection resumed
_
FAQ
Common questions about maintenance windows
Does detection stop during a maintenance window?
No. Detection continues to run and incidents are still opened and recorded in the incident log and audit trail. Only alert delivery is suppressed. This means you have a full record of what happened during maintenance, and you can review the PCAP from any incidents that occurred once the window ends.
What if I forget to end a maintenance window?
Every window has a mandatory maximum duration (set when you create it). If the window isn't ended manually, it ends automatically when that duration expires. There is no way to create an indefinite window. The minimum duration is 5 minutes and the maximum is 24 hours.
Does baseline learning pause during maintenance?
Yes. Samples taken during a maintenance window are excluded from the baseline rolling window. This prevents deploy traffic spikes from inflating the p99 baseline and raising the detection threshold after the window ends.
Can I automate maintenance windows with my CI/CD pipeline?
Yes. The REST API is designed for automation. Add a step at the start of your deploy script to start a window via API, and a step at the end to close it. If your deploy fails and the end step doesn't run, the window ends automatically at its maximum duration. Example integrations for GitHub Actions, GitLab CI, and Jenkins are in the docs.
Related Features