Skip to main content
The health dashboard is a single infrastructure screen, and it collects a summary of all observed entities, shows tables of nodes and connections, keeps a tape of incidents from the last day, and gives a button to immediately start the checkup. Opens in the Health panel.
For whom: panel administrator, SRE, duty engineer. Required rights: Administrator session.

What shows

Card summary

Five cards – nodes, connection schemes (Flow), agents, routes, WARP – broken down by {всего, healthy, degraded, critical, unknown}, as well as counters of open incidents and proposed actions.

Health assessment

Single indicator: the proportion of healthy entities from the total number on all five cards multiplied by 100.

Tables of nodes and diagrams

String state of nodes and connection schemes with filters by status (for nodes - also by country and provider), plus a route table.

Tape of incidents

Health events and synthesized “current issues” over the past 24 hours, each with its own life cycle status.
In addition, the summary separately shows how many problems is hidden by suppressing nodes (hidden_by_suppression) - so that the “green” dashboard is not misleading when part of the nodes is silenced. About the suppression - on the page Node.

Update intervals

Health screen Interrogates the panel every 15 seconds It updates the summary and the tables automatically. This is a reflection of the data already collected: the actual checks are in their intervals (achievable ~60 s, service ~120 s, through circuit check ~10 min - see). Levels of inspectionsTherefore, a fresh result may appear on the dashboard with a delay before the interval of the corresponding check - or immediately if you run the check manually.

The check button (recheck)

On the dashboard there is a global button Run check. It double-checks all the applied connection schemes and returns the number of verified schemes.
For point rechecking, it is not necessary to drive the entire park: an individual circuit or node has its own POST .../check. Manual check makes one through pass and does not escalate to the critical from the node - see Levels of checks.

Incident block

The incident tape combines two things: health events from the magazine and synthesized “current problems” — by circuits, nodes, offline and degraded agents, and by nodes disabled in Remnawave. Each record has a life cycle status: | Status | What is |? |---|---| | open | Problem active, not disassembled by | | acknowledged | Operator has seen / received | | auto_resolved | Target recovered itself, problem closed automatically | | resolved | Closed | The tape can be requested for the desired period and with the inclusion of suppressed records:
Routine tics “still great” in the tape do not get - notifications and incidents are created only on real state transitions. This is noise protection, not data loss.

Chronicle of inspections

The checklist is written for the operator, not for the sampler, and each sample has its own name, and the error text is disassembled according to the meaning: “domain does not resolve”, “port is closed”, “timeout is gone - this is the lock, as opposed to a refusal.” The raw response of the test has not gone away: it lies in the open line verbatim and is signed as raw. Unfamiliar text is shown as it is, because hiding a single specificity for the sake of a beautiful general phrase is the worst place where help is needed. The not_implemented status on the list is called “Unfulfilled”: it is almost always a test that lacked the data to run, rather than a panel defect.

How to read a dashboard

1

First, a global summary.

Is there a critical or degraded on the cards? critical only appears when something user-generated is broken; offline agents themselves raise the state to a maximum of degraded.
2

Then what is it?

Fall into a table with a nonzero critical/degraded and filter by status, so you will see specific nodes and circuits.
3

Then why?

Open the incident or schema card: the cause of the failure, affected nodes/routes and proposed recovery actions are visible.
4

Action.

Recovery actions are offered, but not performed by themselves - see Recovery. For typical scenarios, there are Runbook’ and .

Verification

Everything is green if overall: "healthy", on all cards critical=0 and degraded=0eh open_incidents=0.
At UI, that’s what 100 percent health scores and no red/yellow cards say. If something doesn’t add up, press Run Z and update: maybe the dashboard still shows data until the last run.

Rollback

Dashboard only displays data and runs checks (reading operations) - there is nothing to roll back. Change operations (confirmation of action, suppression) roll back on their pages: Restoration, Suppression nod.

Typical errors

See hidden_by_suppression in the summary: suppression hides events, but doesn’t stop monitoring. Some of the problems may just be hidden.
Recheck double-checks the circuits used and returns the number, and if the circuit is not applied or the node is not applied, there is nothing to change, and the result appears after the run is completed.
If the target is recovered, the record goes to auto_resolved, and the associated action sentences are automatically removed. This is the expected behavior.
Related pages: Conditions and severity, Recovery, Runbook’ and , Notifications, Response to incidents.