Traffic health

Catch a stopped process before your customer does.

odnoga learns the normal traffic of every app and function. It alerts you when a process silently stops or a retry loop drives calls and costs sharply upward — with the history you need to start investigating immediately.

odnoga Functions screen showing a stopped process, a traffic surge, two steady processes and one caller learning its normal — See immediately which process stopped, which one is producing an unusual traffic surge, which ones are steady and which new caller is still learning its normal. Call history and each process’s normal level reveal the scale of the change — without digging through logs.

Five clear states

No score to decode. A state you can act on.

Learning

The caller does not yet have eight comparable windows, so odnoga waits instead of guessing.

Quiet

Its normal volume is below the configured floor, so small changes are not treated as incidents.

Steady

Recent traffic remains inside the caller’s own normal range.

Stopped

Call volume has fallen sharply against the caller’s own baseline.

Surged

Call volume is far above normal — often a retry loop, runaway job or unintended backfill.

From traffic to evidence

Sensitive enough to catch silence. Careful enough not to cry wolf.

Learn each caller’s normal

Every app and function is compared with its own recent windows, not a global average that hides individual behaviour.

Evaluate the current window

odnoga checks whether traffic is steady, too quiet to judge, stopped or far above normal.

Confirm before alerting

A change must persist across consecutive windows. A volume floor, median baseline and cooldown reduce false alarms.

Know where to start looking

The alert includes failed calls recorded in the same window. Open their details to see whether the process stopped sending requests or calls failed somewhere along the delivery path.

Notify and report recovery

Workspace alerts report a confirmed stop or surge, then tell you when normal traffic returns.

Mark intentional changes as expected

An administrator can mark a finished backfill or paused job as expected. Watching resumes automatically when that caller returns.

Where it appears

One finding, the same language everywhere.

Workspace Functions: status, recent-window chart and anomaly count beside the caller.
Organisation Traffic health: every workspace collected in one operational view.
Email: confirmed anomalies and recovery notices go to the workspace alert recipients.
MCP: traffic.health reads the same states and evidence; administrators can mark an intentional change as expected.

Availability depends on your plan. See the live plan comparison for current inclusion and add-on options. See pricing.