An agent on every machine
Linux, macOS and Windows, seven builds. It measures the machine from inside and sends it over HTTPS.
- cpu
- 12 %
- memory
- 41 %
- disk /
- 63 %
- smart
- ok
- Updates itself
- If the network drops, it keeps the batches and sends them later
Server, network and backup monitoring for IT and MSPsSomeone is watching while you sleep.
Sergward watches your servers, your network, your Proxmox nodes, your websites and your backups. When something breaks —or a backup did not run— it tells you, and keeps the history so you can explain what happened.
Free during the beta. Leave your details and we write to give you access.
$ curl -fsSL https://app.sergward.com/install.sh | sudo sh> irm https://app.sergward.com/install.ps1 -OutFile install.ps1; powershell -NoProfile -ExecutionPolicy Bypass -File install.ps1The agent shows you a code and you approve it from the portal. No keys to paste.
From the command to the portal, in a minute
No keys to copy and paste: the agent asks for a code and you approve it from the portal, signed in.
Four ways the data gets in
None of them asks you to open a port into your network.
The same agent watches your network
From inside your network it queries what will not take an agent: switches, UPS units, printers.
- ping
- 11 ms
- loss
- 0 %
- snmp
- port 24 up
- wmi
- ok
- Ping with loss and jitter, and TCP ports
- Scheduled network sweeps
Proxmox, machine by machine
Virtual machines and containers, storage and quorum, the health of each node —ZFS, certificate, subscription— and the last backup of every machine.
- vm 104
- running
- ct 101
- running
- zfs
- online
- backup
- 23 h ago
- With a read-only token
- If a part could not be read, it says which and why
Your websites, from outside
Checks from the platform: the status it must answer, a word that must be there, how long it takes and how many days the certificate has left.
- https
- 200
- response
- 230 ms
- word
- found
- certificate
- 54 d
- Your own method, headers and authentication
- Warns before the certificate expires
A network with no way out to the Internet?
One agent hands its batches to another on the same network that does get out. The one forwarding them does not open them, and the platform takes them only if that permission was given.
Hosts
Every server, inside out
What you would check over SSH or remote desktop, but stored, comparable, and with an alert when it leaves normal.
CPU, memory and load
Total and per-core usage, memory and swap, and the 1, 5 and 15 minute load.
Every disk
Space on every mount point and, from the physical disk, SMART: reallocated sectors, temperature and power-on hours.
Every network card
Traffic in and out per interface, in bits per second.
Processes and services
Which process is eating CPU or memory, and whether the services that matter are running. It finds them itself.
Inventory
Vendor, model, serial number and BIOS, and the installed software with its version.
Linux, macOS and Windows
The same agent in seven builds. On Windows, Windows Server Backup too.
Devices
And everything that will not take an agent
Switches, UPS units, printers, firewalls, NAS boxes and management cards. The agent you already have on the network finds them and queries them over SNMP (v1, v2c and v3), WMI, ping or TCP.
Discovery
Sweeps a subnet and recognises vendor and type from 18 fingerprints: Cisco, HPE iLO, APC, HP printers, Windows, Linux.
Switch ports
Status and traffic of every port, in bits per second.
UPS units
Battery charge, minutes left, output load and whether it is already on battery.
Printers
The level of every toner or supply, before it runs out.
Temperature and hardware
Temperature sensors and, on Cisco, the state of fans and power supplies.
Your vendor's MIB
If the device has something of its own, upload its MIB and pick what to measure.
Alerts that do not make noise
When the gateway goes down you do not get forty emails. You get one, and the rest is stated as its consequence.
Two levels per metric
Warning and critical, for everything, by tag or by machine, plus your own rules.
Dependencies
If the machine others hang from goes down, the ones below stay quiet.
Silence is a failure too
A machine that stops reporting raises an alert instead of leaving a gap in the chart.
Maintenance and on-call
Recurring windows, silencing rules, and who is on duty tonight.
Alerts go out by email, with escalation and a reminder if nobody acknowledges them.
A backup that did not run makes no noise
Until the day you need it. Sergward checks how old each backup is, whether the file came out empty and whether the last attempt failed, and puts it all in one list.
Backup folders
How old the newest file is, how many copies there are and whether any is empty or corrupt.
Windows Server Backup
What Windows itself says about its last run.
Proxmox
Each machine's last good backup, its size, and the machines no job includes.
Dashboards built in three steps
What you want to see, whose, and how. With a live preview on your own data, and changes save themselves.
- 27 widget types and 12 templates offered according to what your machines report.
- One line per machine, per disk, per port or per core.
- A link with no sign-in to put the dashboard on the customer's screen.
- The last 20 versions of each dashboard, to go back.
- PDF reports on availability, capacity and what changed in the period.
If you look after several customers
One organisation with a space per customer, and each one sees only its own.
- Four roles
- Owner, admin, member and viewer, decided per action by the platform.
- Two-factor
- By app or by email, and it can be required for the whole team.
- Audit log
- Who touched what, paginated and exportable to CSV.
- Read-only API
- Keys that only read, with the OpenAPI generated from the real routes.
And what it also does
Less flashy, just as needed the day something breaks.
Inventory
The hardware and software of every machine, with software search, CSV import, and export to CSV and JSON.
Runbook and contacts
Per machine, what to do when it fails and who to call. And tickets with comments.
Sites and rooms
A tree of locations that machines and devices hang from, and tags to cut across them.
Discovery
Scheduled network sweeps, OID walks and, if you switch it on, SNMP trap and syslog reception.
An agent that checks itself
check says where it breaks and doctor fixes what it can on its own. It also has a local dashboard on the machine.
Always the latest version
Two languages, light and dark themes, and it installs as an app. The web updates itself, with no banners.
Pricing
Nothing is charged during the beta. After it, you pay per monitored machine and that is all: every feature is in every plan.
Today, in beta
Everything free, no credit card. Join the waitlist and we write to give you access.
When the beta ends
Free
USD 0 up to 10 machines, forever
- Every feature
- 30 days of history
- No credit card
30 days
USD 2 per machine per month
- Every feature
- 30 days of history
120 days
USD 3 per machine per month
- Every feature
- 120 days of history, to compare against four months
How much would you pay?
30 daysUSD 50per month
120 daysUSD 75per month
Up to 10 machines, free. Above 10, every machine is paid.
- What counts as a machine?
- Every machine with an agent —a server, a PC, a Proxmox node— and every device it watches without one: a switch, a printer or a UPS over SNMP, ping, TCP or WMI. Websites and Proxmox virtual machines do not count.
- Do you charge per user?
- No. Users are not charged: you pay per monitored machine.
- What is history?
- How far back you can look in charts, dashboards and reports: 30 or 120 days depending on the plan.
- Do I have to pay anything now?
- No. Nothing is charged during the beta and no credit card is asked for.
Said plainly
What a buyer asks, answered first.
- 01
It is free while the beta lasts. We give access a few at a time, by hand, from the waitlist.
- 02
Alerts go out by email. No other channels are switched on yet.
- 03
The platform does not ping from outside: pings come from an agent in your network.
- 04
Sergward runs in the cloud. It does not install on your own infrastructure.
- 05
It does not store passwords or give remote control: those are other products.
- 06
Reports are made on demand; they are not sent on a schedule yet.
Join the waitlist
While the beta lasts we give access a few at a time, by hand. Leave your details and we will write to you.
- 1You leave your name and email.
- 2We write to give you access.
- 3You install the first agent.