ARCTIC RELAY MONITOR / OPERATOR GUIDE

Know what
you’re looking at.

A practical reading guide: connect a node, establish a baseline and follow a signal to its source. These examples use the application’s real screens.

← Product overview

FIRST CONNECTION

Verify the server.
Then trust the measurements.

  1. Prepare the connection

    Use Add VPS to enter the Ubuntu server connection and select the SSH key. Keep account details in the application, never in a public screenshot.

  2. Verify the host identity

    Compare the displayed SSH fingerprint with the provider console or another trusted source before accepting it. Fetching a key alone does not authenticate the server.

  3. Review and connect

    Run the read-only checks. Review and approve any required agent installation separately, then wait for a fresh, authenticated sample.

  4. Establish a baseline

    Open Details, confirm the connection checks and review CPU, workload RAM, disk and peer readings. Return to Fleet for the broader picture.

FOUR DISTINCTIONS THAT MATTER

Read the source, not just the number.

Mac RTT / Peer RTT
Mac RTT describes the operator’s connection to the VPS. Peer RTT describes measurements between VPS neighbours. A problem on one path need not affect the other.
Availability / SLA
Availability reflects monitoring observations from this Mac. A displayed 100% is not an independent provider SLA measurement.
Zero / Not configured
A zero and an unavailable source are different states. Check the source status before interpreting queues, throughput or TLS fields.
Health / Capacity
Health describes current findings. Capacity estimates use local history to help plan ahead; they do not purchase or scale servers.

01 / Fleet / expanded view

Read one card before opening its details

Expanded cards expose resource utilisation, uptime, peer RTT, Mac RTT, outbound queue and network rates directly in the fleet view.

Fleet / expanded view
An empty TLS field represents missing telemetry. It should not be interpreted as a failed certificate check.

When to use it

Choose this layout when a few nodes need closer attention. Use Details for the selected node; change back to compact mode when the whole fleet matters more.

What to read

  • Uptime and resource bars
  • Peer RTT versus Mac RTT
  • Queue and network-rate context

02 / Node details / advanced telemetry

Separate memory pressure from memory accounting

Workload RAM, allocatable memory, provider balloon and memory pressure appear beside optional process, file-descriptor, SQLite and network counters.

Node details / advanced telemetry
Awaiting Relay and Not configured are visible in this capture. Those fields require their data source; enabling a display checkbox does not create the missing measurements.

When to use it

When RAM looks high, check workload and pressure first. Provider ballooning alone is not evidence of an application memory leak. The checkboxes control which metrics are displayed.

What to read

  • Workload, allocatable RAM and PSI
  • Optional Relay process and SQLite data
  • RX/TX totals and error growth

03 / Node details / capacity and history

Wait for a useful trend

Resource summaries sit above storage, RAM and connection forecasts, followed by disk, latency, availability and queue history.

Node details / capacity and history
This capture contains only 0.1 hours of samples and shows No forecast. The displayed minimum is 12 samples over at least one hour; an unstable trend is also withheld.

When to use it

Use the history to distinguish a short spike from sustained growth. Read a forecast together with the amount and consistency of the data behind it.

What to read

  • Average and peak resource use
  • Local trend estimates
  • Availability and queue history

04 / Node details / throughput

Distinguish server traffic from Relay activity

The network graph shows inbound and outbound server traffic while the Relay graph remains flat. Source status below the graphs explains why these measurements differ.

Node details / throughput
Here Relay is Not configured and TLS is Not monitored. The Disk I/O R/W value Agent v1 is not a disk-speed measurement. The guide also treats Relayed today as a cumulative counter rather than a verified daily total.

When to use it

Confirm that the required source is configured before interpreting a zero. Network bytes can change even when there is no configured Relay telemetry.

What to read

  • Inbound and outbound network rates
  • Relay counters when configured
  • Agent version and telemetry status

05 / Compare / compact cards

Keep several dimensions beside each other

A compact comparison places CPU, memory, disk, swap, load, availability, latency and error rates on each node card. Provider and region labels support the investigation.

Compare / compact cards
A highlighted network-error rate is a reason to investigate its source. It does not by itself establish an attack or determine the node health score.

When to use it

Look for a shared pattern across nodes before narrowing to a single server. Use a common period so the values are comparable.

What to read

  • Resources, load and swap
  • Peer and monitoring latency
  • Provider and region context

06 / Timeline / local event journal

Follow problems and recoveries through time

The horizontal axis is time and the vertical axis is severity. Filters select problems, recoveries, capacity events and renewals for a fleet or one VPS.

Timeline / local event journal
No events match this captured 24-hour range. The empty state is deliberate: it does not manufacture activity or prove the absence of incidents outside the selected journal and interval.

When to use it

Start with a short period around an incident, add recovery events, then focus on the relevant node. In the application, hovering explains an event and clicking keeps it selected.

What to read

  • Time and severity
  • Problems, recoveries and planned events
  • Per-node and fleet filters

A REPEATABLE OPERATOR ROUTINE

Scan. Investigate. Review.

DAILY

Read the current state

Check Fleet and sample freshness. Open nodes that need attention. Review memory pressure, disk, configured Relay counters and slow peer links.

AFTER A CHANGE

Confirm the recovery

Check a fresh sample, compare the affected metrics and inspect the incident history. End maintenance when the planned work is complete.

WEEKLY

Look beyond a single spike

Compare a longer period and its previous window. Review recurring incidents and capacity trends before deciding what needs intervention.

READING THESE EXAMPLES

Keep the limits visible.

The screenshots show a particular session, including unconfigured Relay fields and an empty Timeline. They demonstrate the interface and the available evidence, without substituting invented activity.

These explanations draw on the Arctic Relay Monitor operator manual and the current application’s documented behaviour. Features that depend on optional telemetry remain dependent on that source. This concise guide does not replace the application’s confirmation steps or server-access requirements.

Back to Monitor overview ↗

Arctic Relay Monitor

Zoom in for small labels. Scroll across the screen to inspect details.