InstallWhat the colours meanWhat the alerts sayReports and client pagesThe config fileSomething is wrongUninstallBilling and cancelling
One line per server. Your account page shows these with your key already filled in; here the key is YOUR_API_KEY.
curl -fsSL https://monitor.servasight.com/install.sh | sudo bash -s -- YOUR_API_KEY
Needs Python 3 and cron, which every server distribution has. Puts the agent in /opt/servasight-agent, the settings in /etc/servasight-agent/agent.conf, a job in /etc/cron.d/servasight-agent that runs every 5 minutes, and sends the first report right away. Log: /var/log/servasight-agent.log.
Open PowerShell as Administrator and paste:
powershell -ExecutionPolicy Bypass -Command "iwr -useb https://monitor.servasight.com/install.ps1 -OutFile $env:TEMP\install.ps1; & $env:TEMP\install.ps1 -ApiKey YOUR_API_KEY"
Nothing else to install. Puts servasight-agent.exe and the settings in C:\ProgramData\servasight-agent, creates a scheduled task named Servasight Agent that runs every 5 minutes as SYSTEM, and sends the first report right away.
The server appears on your dashboard within a minute. If it does not, see Something is wrong.
The agent reads RAID, power supply and temperature from the vendor tool when one is installed: HPE ssacli and hpasmcli, Dell perccli / storcli and omreport, any BMC through ipmitool, and lm-sensors on Linux. On Windows it also reads Get-PhysicalDisk. When no tool is present that column says so and everything else still works.
| Green | Yellow | Red | |
|---|---|---|---|
| Disk space | under 85% full | 85% or more | 95% or more |
| Drives and RAID | every array and drive OK | rebuilding, predictive failure, or no RAID tool found | degraded or failed |
| Power and temperature | all supplies OK, temperatures normal | a warning reading, or no hardware tool found | a failed supply or a critical temperature |
| Reboots and silence | reporting every 5 minutes | unexpected reboot, or silent for 20 minutes | silent for 60 minutes |
| Warranty | more than 60 days left | 60 days or fewer | expired |
| Services and ports | every listed service running, every listed port answering | a listed service is not installed on that server, or still starting | a listed service is stopped, or a port does not answer |
| Clock | within a minute of the time servers | a minute or more off, or updates waiting for a reboot (shown on Uptime) | five minutes or more off |
| Backups | every listed backup newer than its limit (26 hours unless you say otherwise) | older than the limit | older than twice the limit, or no backup found at all |
| Updates | patched within 45 days | 45 days or more since the last patch, or the date is unknown | 90 days or more |
| Certificates and domains | certificates good for 30+ days, domains for 60+ | a certificate under 30 days, a domain under 60, or one that could not be checked | a certificate under 7 days or expired, a domain under 14 |
A server's overall colour is its worst column. Yellow means look at it this week. Red means look at it now.
FIXED: the agent put something right on its own: restarted a stopped service you listed, thinned a filling disk, or resynced a clock that had drifted (whichever of auto_fix = services, disk, clock you turned on). The message says what it restarted and what the outage would have meant for that office, in plain words. The dashboard keeps the last ten under "Fixed automatically".
You get one message when a problem starts and one when it clears. Nothing repeats while the problem stays the same. Alerts go to the contacts on file for your account; to add a phone number, an email address or a Pushover key, reply to your welcome email and we set it the same day.
Monthly report (every plan): on your account page, each client has a Monthly report link: one printable page for the month, with your logo and footer line, showing uptime, what was caught, what was fixed automatically, disk trend, backups, patching, certificates and warranty. Use the browser's print button to save it as a PDF and send it with the invoice.
Status page and share link (Pro): each client also gets a public status page (a green, yellow or red dot per server, nothing else) and a share link to the report itself. Hand the links to the client; nothing on them needs a login.
Monthly email (Pro): put the client's email on the account page and the report summary plus the share link goes out by itself on the 1st of each month.
Monday digest (every plan): one message every Monday morning to your alert contacts: servers watched, problems caught, fixed automatically, and what is trending toward trouble (disks over 85%, stale backups, certificates under 30 days, patching overdue, warranty under 60 days).
Scheduled reboot (Pro): when a server says updates are waiting for a reboot, its row on the dashboard shows Reboot tonight at 02:00. One click, and the agent restarts that server between 02:00 and 05:00 server time, once, only if the OS still asks for it, and tells you it did.
Linux: /etc/servasight-agent/agent.conf. Windows: C:\ProgramData\servasight-agent\agent.conf. Plain text, one setting per line. The installer fills in the first three.
| server_url | Where reports go. Leave it alone. |
|---|---|
| api_key | Your key from the account page. One key for all your servers. |
| server_name | The name shown on the dashboard. Change it to anything, for example Smile-Dental-FS01. |
| client | Whose server this is, for example Smile Dental. The dashboard groups servers by client and counts what was caught and fixed for each one this month. You can also give it on the install line: Linux ... YOUR_KEY "" "Smile Dental", Windows -Client "Smile Dental". |
| warranty_end | YYYY-MM-DD. Blank means the warranty column stays grey. |
| disk_warn_percent / disk_crit_percent | Yellow and red lines for disks. Default 85 and 95. |
| warranty_warn_days | How many days before the warranty ends to go yellow. Default 60. |
| mounts | Only watch these mount points or drive letters, comma separated. Blank means every real disk. |
| services | Services that must be running, comma separated. Windows: the service name or its display name, for example MSSQLSERVER, Print Spooler. Linux: the systemd unit, for example postgresql, smbd. Blank means the Services column says "none listed". |
| auto_fix | What the agent may put right on its own, comma separated; blank means report only, touch nothing. services: start a listed service again the moment it is found stopped (3 times a day). disk: when a disk reaches the yellow line, delete temp files older than 7 days, rotated logs older than 30 days, package caches and old update downloads, never your data or the recycle bin (once a day per disk). clock: resync a clock a minute or more off (3 times a day). spooler (Windows): clear print jobs stuck in error and restart the spooler (3 times a day). hung: a listed service stuck starting or stopping for 10 minutes is killed and started clean (3 times a day). Every fix sends one FIXED message with what was done and what it would have meant. |
| backups | Backups that must be fresh, comma separated, label=where. where is a file pattern (Veeam=D:\Backups\*.vbk, Files=/srv/backup/*.tar.gz) or wbadmin for Windows Server Backup. Add @hours to one item for its own age limit (Weekly=E:\wk\*.bak@170). |
| backup_max_age_hours | How old a backup may be before it goes yellow. Default 26. Twice that is red. |
| certs | Hosts whose TLS certificate must not expire, comma separated, host or host:port: www.smiledental.com, mail.smiledental.com:993. Self-signed and internal certificates are read too. |
| domains | Domain names whose renewal date is watched (checked once a day): smiledental.com. |
| ports | TCP ports that must answer, comma separated. A bare port means this server, host:port checks another machine from here, for example 1433, 445, 192.168.1.10:443. |
Changes take effect on the next 5-minute run. No restart needed.
sudo python3 /opt/servasight-agent/servasight_agent.pyC:\ProgramData\servasight-agent\servasight-agent.execat /etc/cron.d/servasight-agent. Windows: Task Scheduler, look for Servasight Agent.The agent is not getting through. Same three checks as above. The colour clears by itself on the next successful report.
Install the vendor tool on that server (ssacli or hpasmcli for HPE, perccli or omreport for Dell, ipmitool for any BMC). The agent picks it up on the next run.
The name in services does not match a service on that server. Windows: open Services, right-click the service, Properties, and copy the Service name (or the display name exactly as shown). Linux: systemctl list-units --type=service and copy the unit name without .service.
Reports are refused until payment goes through. Open your account page and complete payment or update the card.
Linux:
sudo rm -f /etc/cron.d/servasight-agent sudo rm -rf /opt/servasight-agent /etc/servasight-agent
Windows (PowerShell as Administrator):
schtasks /Delete /F /TN "Servasight Agent" Remove-Item -Recurse -Force "$env:ProgramData\servasight-agent"
The server goes yellow after 20 minutes and red after 60, then stays red until you remove it or it reports again. Removing a server from the dashboard: reply to your welcome email with its name.
| Starter | $249 a month | Up to 10 servers, alerts to two people. Every check: disk, RAID, power and heat, reboots, warranty, services, clock, backups, patching, certificates and domains. Auto-fix for stopped services, filling disks, drifted clocks, stuck print queues and hung services. Dashboard by client, a printable monthly report per client with your logo, the Monday digest. |
|---|---|---|
| Pro | $399 a month | Up to 50 servers, alerts to five people. Everything in Starter, plus a public status page per client, the monthly report emailed to each client automatically with a share link, and after-hours reboots you schedule with one click. |
| Unlimited | $599 a month | No limits: unlimited servers, unlimited alert contacts. Everything in Pro, plus white-label reports and status pages (your name only), a JSON feed of every server for your own tools, and priority support. |
Per company, never per device. Going over your plan's server count shows a notice on the dashboard and account page; use Manage billing to move up.
$249 / month per company (Starter, up to 10 servers). Billing is handled by Stripe. The Manage billing button on your account page opens Stripe's portal where you can change the card, download invoices or cancel.
Cancelling stops monitoring at the end of the month you have paid for. No contract, no cancellation fee. A failed card keeps monitoring on while Stripe retries for a few days, then turns it off until the payment succeeds.