A monitoring tool only earns its keep after four things are done properly — credentials, discovery, monitors and alarms. This guide walks the complete OpManager setup in that order, with the honest checks that tell you it is really working.
Most OpManager installs look finished long before they are. The installer completes, the dashboard opens, a few devices appear — and the admin moves on. Then the first switch dies at midnight and nobody gets an alert, because the credentials were never tested, the device was discovered but never approved, or the alarm had no notification profile behind it. OpManager is genuinely quick to set up — experienced admins have it scanning a subnet within about twenty minutes of install — but the speed hides the four load-bearing steps. They are: credentials first (SNMP, WMI or CLI, added once and reused), discovery second (an IP range scan whose results you approve into the inventory), monitors third (CPU, memory, disk and interface monitors that attach automatically to SNMP devices — and must be verified on the device snapshot page), and alarms last (thresholds plus a notification profile, tested with one real alert). Do them in this order and the video walkthrough above becomes a checklist instead of a hope.
The order is the whole secret. Alarms configured before discovery are alarms attached to nothing.
Example values. The snapshot page is your ground truth — if the dials move, the monitor works.
Download the installer only from ManageEngine’s official site and install it on a Windows or Linux server that stays powered on — a monitoring server that sleeps at night monitors nothing. OpManager runs as a web application; after install you reach the console in a browser on its default web port 8060. Create your admin account properly at first login and note two sizing truths: the server needs a static IP or a fixed DNS name (probes, devices and your bookmarks all point at it), and its clock must be right, because every graph and alarm timestamp inherits it.
OpManager talks to devices through SNMP, WMI and CLI (SSH/Telnet), and it stores the logins centrally so one credential can serve a whole fleet. Skipping this step is the single most common setup mistake — discovery then finds boxes it cannot actually read.
DOMAIN\username plus its password161 and the suggested timeout of about 5 seconds unless your WAN links are slowOpManager uses ICMP and Nmap-style probing to find devices. Right after install it already picks up neighbouring servers and domain controllers; for the rest of the network you run a proper discovery. Newer builds place this under Settings → Network Discovery; the flow is the same everywhere.
For every SNMP-enabled device, OpManager automatically associates CPU, memory and disk monitors, and interface monitors for its ports. Do not trust the automation blindly — verify one server and one switch by hand. Open the device’s snapshot page. If SNMP is really flowing, the sysDescr field shows the device’s own description and the resource dials move. Use the Test Monitor link against any monitor that looks dead; it queries the device live. A device that answers ping but shows no sysDescr is only availability-monitored — useful, but it is not performance monitoring, and the fix is always on the credential or the device’s SNMP settings, never in the graph.
Alarms are the product. A threshold says when a monitor raises one (CPU above 90 percent, a WAN interface above 85 percent utilisation, a disk below 10 percent free), and a notification profile says who hears about it — email, SMS or a ticket in your helpdesk. Configure both, then prove the chain end to end: pick one unimportant device, set its CPU threshold absurdly low for five minutes, and watch the alarm arrive in your inbox. That five-minute test is the difference between a monitoring system and a monitoring installation. The companion pages below cover the failure modes in depth — discoveries that find nothing, alarm storms, WAN links, custom monitors, editions and backups.
Installation, credential setup, discovery and the first dashboard are all shown live in the OpManager setup video — watch it once with your own console open beside it.
▶️ Watch Full VideoNo. OpManager is agentless — it polls devices over SNMP, WMI or CLI using the credentials you store under Settings → Discovery → Device Credentials. That is why the credential step matters so much: no agent also means no data at all when a credential is wrong.
Almost always because SNMP is not really answering on that device. Open its snapshot page: if sysDescr is blank, OpManager only has ping-level availability. Enable SNMP on the device, match the credential, and use Test Monitor to confirm.
By default the console is served on port 8060 over HTTP, and it can be switched to HTTPS with your own certificate. Keep that port reachable for the admins who need the console — and not open to the whole internet.