Server monitoring that installs nothing

Live status for every machine, over the SSH connection you already use. Add a host, and the first card fills in within seconds.

Total

Online

Offline

🇦🇺Web Gateway15 ms

8 Cores16.0 G6 d

CPU

36%

RAM

47%

Network

5.21 M/s10.30 M/s

Disk

13.5 M/s12.0 M/s

🇺🇸Build Runner29 ms

12 Cores16.0 G7 d 1 h

CPU

54%

RAM

49%

Network

5.67 M/s8.78 M/s

Disk

15.4 M/s14.5 M/s

🇩🇪Database Node30 ms

8 Cores16.0 G8 d 2 h

CPU

26%

RAM

54%

Network

5.20 M/s8.92 M/s

Disk

16.2 M/s17.1 M/s

🇯🇵Edge Cache42 ms

4 Cores8.0 G3 d 4 h

CPU

19%

RAM

39%

Network

3.14 M/s6.43 M/s

Disk

8.9 M/s9.6 M/s

One machine, in full

Each server gets a stack of cards, from processor to disks. A card appears only when the machine has something to show.

Drag the cards into your order, and hide the rest.
One layout for every machine.

Every card on the status page

Each card answers one question about the machine.

Every core, broken down

Overall usage with the CPU model, and a row per core split into user, system, I/O wait, and more.

The shape of the load

The one, five, and fifteen minute averages on one chart, so a spike and a slow climb look different.

Memory that is really in use

Memory in use is shown apart from cache Linux can reclaim, so 90 percent stops looking alarming.

Find what is eating the machine

A live table, heaviest first and sortable by any column. Open it to search by name, user, or PID.

What the interfaces are doing

Live throughput and totals for the busiest interface. With vnStat, history reaches back years.

A full disk and a busy disk

Capacity for every volume, and speed, latency, and IOPS for every disk.

Where the server really is

Public IP, country, city, organization, and a map pin. The lookup runs from your server to ipinfo.io, and the result stays on your device.

Containers, beside everything else

A tile per container with CPU, memory, network, and disk I/O. The full Docker panel is one tap away.

Live NVIDIA card status

Utilization, VRAM, temperature, and power for every NVIDIA card, plus the processes using it.

9:41

SwiftServer

Total

Online

Offline

🇦🇺Web Gateway15 ms

8 Cores16.0 G6 d

CPU

36%

RAM

47%

Network

5.21 M/s10.30 M/s

Disk

13.5 M/s12.0 M/s

🇺🇸Build Runner29 ms

12 Cores16.0 G7 d 1 h

CPU

54%

RAM

49%

Network

5.67 M/s8.78 M/s

Disk

15.4 M/s14.5 M/s

🇩🇪Database Node30 ms

8 Cores16.0 G8 d 2 h

CPU

26%

RAM

54%

Network

5.20 M/s8.92 M/s

Disk

16.2 M/s17.1 M/s

🇯🇵Edge Cache42 ms

4 Cores8.0 G3 d 4 h

CPU

19%

RAM

39%

Network

3.14 M/s6.43 M/s

Disk

8.9 M/s9.6 M/s

Web Gateway

🇦🇺Web Gateway15 ms

8 Cores16.0 G6 d

CPU

36%

RAM

47%

Network

5.21 M/s10.30 M/s

Disk

13.5 M/s12.0 M/s

Web GatewayUbuntu 24.04 LTS

CPU Usage44°C

37%SwiftServer Demo Compute

60%

25%

22%

17%

57%

50%

41%

Cores8

User

System

Nice

IOWait

Steal

CPU Load

18:12

18:42

19:12

1m5m15m

Current Load1.27 / 0.88 / 1.45

Memory Usage

Total16.0G

7.36G

2.88G

5.76G

Swap696 M of 2.00 G Used

34%66%

Processes

PidProcessUserCPU%Mem

1866swiftserver-webdeploy10.8358 M

1903swiftserver-apideploy9.8433 M

2125prometheusmonitor6.0452 M

1940swiftserver-workerdeploy4.8322 M

2014nginxwww-data4.1157 M

2051envoyproxy3.3281 M

1977nginxwww-data3.2162 M

2236containerdroot3.2206 M

1644vectormonitor2.5153 M

2347healthcheckdeploy2.1126 M

1533tailscaledroot1.998.7 M

2162grafana-servergrafana1.8323 M

2199dockerdroot1.7304 M

Network Usage

eth0192.168.10.24/24

4.01M/s8.39M/s

24.2G78.2G

vnStat

HourDayWeekMonthYear

Upload7.90 MPast Hour

8:00 AM12:00 PM4:00 PM

Download33.3 MPast Hour

8:00 AM12:00 PM4:00 PM

Storage

/dev/vda126.1G / 112G

/ext4

SpeedLatencyIOPSOverall

8.39M/s0.42ms214590G

4.01M/s1.20ms3881.72T

IP Location

🇦🇺 Australia192.168.10.24

CityMelbourne, Victoria

OrganizationAS64500 SwiftServer Demo Network

Docker

Running containers3

demo-api3 hours

CPURAMNetworkBlock IO

18%

128M42.0M42.8M16.2M

demo-worker2 hours

CPURAMNetworkBlock IO

18%

128M42.0M42.8M16.2M

demo-cache3 hours

CPURAMNetworkBlock IO

18%

128M42.0M42.8M16.2M

demo-dbExited (0) 4 minutes ago

CPURAMNetworkBlock IO

0B0B0B0B

demo-grafanaExited (0) 21 minutes ago

CPURAMNetworkBlock IO

0B0B0B0B

GPU Usage

Driver: 550.54CUDA: 12.4

GPU: 0NVIDIA Demo GPU

UtilMemFanTemp

62%

48%

41%

58°C

Memory Free12600 M / 24576 M

Power168 W / 320 W

Processes Running2

PidProcessMem

1801python6144 M

1812worker2048 M

DashboardSessionsResourcesSettings

One machine, in full

Each server gets a stack of cards, from processor to disks. A card appears only when the machine has something to show.

Drag the cards into your order, and hide the rest.
One layout for every machine.

Every card on the status page

Each card answers one question about the machine.

Every core, broken down

Overall usage with the CPU model, and a row per core split into user, system, I/O wait, and more.

The shape of the load

The one, five, and fifteen minute averages on one chart, so a spike and a slow climb look different.

Memory that is really in use

Memory in use is shown apart from cache Linux can reclaim, so 90 percent stops looking alarming.

Find what is eating the machine

A live table, heaviest first and sortable by any column. Open it to search by name, user, or PID.

What the interfaces are doing

Live throughput and totals for the busiest interface. With vnStat, history reaches back years.

A full disk and a busy disk

Capacity for every volume, and speed, latency, and IOPS for every disk.

Where the server really is

Public IP, country, city, organization, and a map pin. The lookup runs from your server to ipinfo.io, and the result stays on your device.

Containers, beside everything else

A tile per container with CPU, memory, network, and disk I/O. The full Docker panel is one tap away.

Live NVIDIA card status

Utilization, VRAM, temperature, and power for every NVIDIA card, plus the processes using it.

Details

Read-only

Uses the tools already on the server, and leaves nothing running.

Linux, macOS, Windows

Each system is read its own way, and fills in on the first connection.

Your refresh rate

Set it for each machine, from once a second to once a minute.

On the Home Screen

Widgets show which servers are online, or one machine in full. Live Activities keep sessions on the Lock Screen.

Every server you run

SwiftServer recognizes each system on the first connection and reads it the right way.

Linux

Ubuntu
Debian
Fedora
Rocky Linux
AlmaLinux
CentOS
Arch Linux
Alpine
openSUSE
NixOS
Kali
Any other distribution

Home lab

Raspberry Pi
Proxmox VE
OpenWrt routers

macOS and Windows

macOS
Windows 10 and 11
Windows Server

How it works

Add a machine

Host, port, user, and a password, an SSH key, or a saved Identity. Keyboard-interactive prompts are supported as well.

Trust the host key

The first connection shows the endpoint, the SHA256 fingerprint, and the server banner. Trust it once and it is remembered.

Turn on Show Status in Dashboard

This is what makes a machine appear as a monitored card. The free tier monitors three at a time, and saved hosts are unlimited.

SwiftServer detects the OS

It identifies the distribution or operating system and reads the machine the way that system expects.

Metrics start arriving

The cards fill in within seconds. Open one for the full chart, the process table, or the breakdown by interface.

Monitoring docs

Questions

Do I need to install an agent to monitor a server?

No. SwiftServer runs read-only commands over the SSH connection you already use and reads what comes back. There is nothing to install on the server and nothing new listening on a port.

Which operating systems can SwiftServer monitor?

Linux, macOS, and Windows. Linux has the full card set. macOS has no vnStat traffic history. Windows has no Docker panel, load average, or temperature, but it does include NVIDIA GPU.

Can it monitor a Raspberry Pi, a Proxmox host, or an OpenWrt router?

Yes. Anything that runs Linux and accepts SSH gets the Linux card set, including Raspberry Pi OS, Proxmox VE hosts, and OpenWrt routers with their BusyBox userland. There is nothing to install on them.

How often does it refresh?

The dashboard refresh defaults to 5 seconds and can be set anywhere from 1 to 60 seconds. Longer-term readings, such as traffic history, refresh on their own slower schedule.

Does SwiftServer send alerts when a server goes down?

No. There are no threshold alerts and no metric notifications. SwiftServer shows state while you are looking, in the app and in the widgets.

How many servers can I monitor for free?

Three at a time on the dashboard. The number of hosts you can save, connect to, and open a terminal on is unlimited. Premium removes the three-server monitoring limit.

Does it keep a history of my metrics?

Metric history stays in memory while the app is open. Network traffic history comes from vnStat on the server, which is why it survives a restart.

Keep reading

GPU monitoringUtilization, VRAM, temperature, power, and the processes holding each card.

DockerContainers, Compose, images, volumes, networks, and live daemon events.

TerminalA complete xterm-256color shell on a first-party SSH engine.

Every server, at a glance.

Free on iPhone, iPad, and Mac.