Getting Started

Set Up Your Grid

Follow the guided tour from an empty grid to measured health, alerts and live displays.

Open the guided tour

Every new grid opens a Guided Tour in Design mode. Use Back and Next at your own pace. The buttons open the relevant editor; they never enroll an agent, create a connection or run a command without your input. Read The Guide opens the full instructions for that topic.

Choose Skip Tour or close the card to dismiss it for this grid. Guided Tour in the top toolbar restarts it from the beginning. All Steps lets you jump to a topic. Progress and dismissal are remembered separately for each grid in this browser. A different browser or cleared browser storage may show it again. Observe hides the tour; return to Design to continue. Organization-administrator topics are omitted for users without that access.

1. Name and map your systems

After registering, enter your Organization Name and name your first grid. Add more grids from the grid-name menu. Start with one rack, one server or one signal path rather than your entire estate.

Drag a template from Equipment, or choose a starting layout in the tour. Give equipment its actual name. Edit its interfaces when the template does not match. Custom equipment can be scoped to one grid or shared across the organization’s grids. Large switches use compact port matrices.

Read Using The Grid, Equipment And Physical Ports and Grids And Dashboards.

2. Choose how readings reach Sysavia

What you needStart here
CPU, memory, disk and services on one hostInstall an agent with This Machine or Both
Private switches, UPSs and equipment APIsInstall a This Network or Both agent and permit their exact addresses
Readings from an equipment APIAdd a System Connection, discover/import its specification, then select a reading
Switch interfaces and UPS readingsAdd an SNMP connection with the device’s credentials
Hardware power and sensorsAdd Redfish or IPMI through an agent that can reach the controller
Remote commands on an SSH deviceSave an approved operation and verify its host fingerprint

Follow Install The Native Agent and Connect Your First System. One agent can reach many devices, but each still needs an accessible management interface and suitable credentials. Discovered operations are suggestions to review, not proof of write permission.

3. Verify one measurement

Test the connection, choose a reading, then use Monitor This Reading to attach it to equipment. Confirm the observed value changes when the real system changes. A green diagram port alone does not verify a cable or service. Diagram Only, Setup Required and Reporting Stale tell you what remains unverified.

For TCP services, use a TCP Port check; for physical interface state, use SNMP, an API or traps. Monitor Network Ports explains the difference and why a timeout does not prove a firewall port is closed.

4. Add wires, metrics and labels

Click a source port followed by its destination. Match direction, transport and connector. Choose the wire’s route style and label location. A drawn wire documents a relationship; a measured check supplies health.

Use + Metrics to place one or several readings, status pills and optional graphs. Name each reading for the operator who will use it. Configure text and shapes to show variables or change color with state. See Metric Panels And History and Live Values And Status Colors.

5. Add an alert, then recovery

Create an automation with a specific check or event, conditions and an alert action. Set matching cycles and cooldown so brief interruptions do not flood Activity. Verify the trigger before adding a real restart, port change, script or file transfer.

For recovery, set a fresh verification check, timeout and failure action. An accepted command is not proof the device recovered. Scripts must be explicitly allowed on their execution agent; file jobs refer to named transfers configured in Agents. See Build And Verify An Automation and Scripts And Monitoring Recipes.

6. Add optional network access and device events

For remote sites, VPNs And Network Tools explains selecting a Linux execution agent, importing a private-route profile and checking its state. Host VPN tools and privileges are prerequisites; configuration and operation then happen in the panel.

For event-driven equipment, Receiving SNMP Traps explains the listener, source allowlist, authentication and mappings. The equipment must send traps to the agent’s reachable UDP address. Configure a recovery event and retain a polling check; silence is not recovery.

For out-of-band management, use Redfish And IPMI Controllers. Keep power control permissions separate from observation.

7. Add a live display

Create and test a media source, then add its playback window to this or another grid. Move and resize it independently of the equipment. Playback pauses when its grid is hidden.

Private VNC requires an enabled agent console relay and a permitted target. IPMI itself is not a universal video protocol. Proprietary controller launch pages may need browser access or a supported gateway. See Media Sources And Playback Windows and Remote KVM And VM Consoles.

8. Test, share and observe

Use Simulate for modeled failures. It does not prove the physical connection or recovery command works. Test actual monitoring and actions in a controlled maintenance window before relying on them.

Give users only the grid permissions they need. Verify their view before handing over the dashboard. See Organizations And Grid Permissions.

Observe shows live changes and hides editing panels. Escape returns to editing. Set automatic grid rotation and alert following when useful; Activity remains shared across grids you can access. You do not need every optional integration to have a useful grid.

When setup stalls

  • Agent offline: verify the service is running, its host can reach the organization’s HTTPS address, and enrollment has not expired or been revoked.
  • Configuration applying: wait for the next heartbeat; verify the agent is online and check its configuration error.
  • Device unreachable: check the chosen agent, permitted address, route, firewall and device credentials.
  • No discovered API: import the manufacturer’s specification or configure a documented operation. A login page is not enough to infer controls safely.
  • Command accepted but still unhealthy: inspect the command result and a new health reading; use the automation’s failure path.
  • No feed: test the source before placing playback, then check the execution host, gateway, authentication and protocol requirements in its guide.