Zabbix graphs, dashboards and network maps

Everything happens in the frontend at https://zbx01.lab as Admin, except the CPU load in step 9, which runs on web01. web01 must be linked to Linux by Zabbix agent active and in the host group Lab/Web, as in Lessons 4 and 6. The Zabbix server’s own host is called zbx01 here and sits in the built-in group Zabbix servers; if you kept the default name “Zabbix server”, read zbx01 as that.

1Look at a simple graph

Open Monitoring > Latest data, type web01 in the Hosts filter and press Apply. In the Name filter type CPU utilization, then click the Graph link at the end of the row.

The graph opens for the last hour with one line, and the legend under it lists CPU utilization with its last, min, avg and max values. On an idle web01 the line stays at a few percent.

That’s a simple graph, made on the fly from history. Change the time selector at the top to Last 7 days and the line gets smoother, because periods longer than the history window and big ranges read hourly trends. Tick two items in Latest data and use Display graph at the bottom to see them together.

✅ Checkpoint: You opened a simple graph for web01 CPU utilization and changed its time period.

2Build a custom stacked graph on web01

Go to Data collection > Hosts, click Graphs in the web01 row, then Create graph. Fill in the form like this and press Add:

Now open Monitoring > Hosts, click Graphs in the web01 row and pick the new graph from the list.

The graph shows three coloured bands stacked on top of each other: green for user time, red for system time and blue for iowait. The legend under it lists each item with its values, and on an idle web01 the top edge stays at a few percent.

Stacked means each item sits on top of the one below it, so the top edge is the total. This graph is saved on web01 only; in Lesson 10 you’ll move graphs like this into a template so every web server gets one.

✅ Checkpoint: The graph “web01: CPU user vs system” shows three stacked colours under Monitoring > Hosts > Graphs.

3Create the “Lab NOC” global dashboard

Click Dashboards in the menu, then the All dashboards link at the top, then Create dashboard. In the Dashboard properties window set Name to Lab NOC, leave Owner as Admin and Default slideshow interval at 30 seconds, and press Apply. You’re now in edit mode with an empty grid.

Click Add widget (or click an empty spot on the grid), choose type Problems, name it Open problems, put Lab/Web and Zabbix servers in Host groups, tick Show operational data: With problem name and press Add. Stretch it across the top of the grid.

The widget titled Open problems shows the text No data found.

“No data found” is good news here: no open problems on those groups. If you still have problems open from Lesson 8, you’ll see them listed with severity colours.

✅ Checkpoint: The dashboard Lab NOC exists in edit mode with a Problems widget across the top.

4Add a Graph widget with patterns

Add another widget, type Graph, name CPU, all lab hosts. On the Data set tab type * in Host patterns (or pick web01 and zbx01) and CPU utilization in Item patterns. On the Displaying options tab leave History data selection at Auto, and on Legend tick Display min/avg/max.

Press Add.

The widget draws one line per host, web01 and zbx01, and because you ticked Display min/avg/max, the legend shows min, avg and max next to each host’s CPU utilization.

Because you used patterns, a new Linux host with the same item name will appear on this graph by itself. If zbx01 is missing, check that the zbx01 host has the item “CPU utilization” (it comes from the Linux template linked to it).

✅ Checkpoint: The Graph widget draws one line per host for CPU utilization.

5Add Top hosts, Item value, Gauge, Honeycomb and Clock

Add these five widgets the same way. Keep each small and arrange them in a row under the graph.

Press Save changes at the top right. Edits are not stored until you do this, and closing the tab before saving throws them away.

✅ Checkpoint: Lab NOC has seven widgets and you saved it.

6Add a second page and share the dashboard

Click Edit dashboard, then the arrow next to Add and choose Add page. Name it Maps and press Apply; you’ll put the map here in step 8. Save changes, then open the Actions menu (top right) and choose Sharing.

Before you share, create a viewer to share with. Lesson 18 covers users properly, so keep this one simple:

Back in Sharing, leave Type as Private, add “Lab viewers” to the List of user group shares with Read-only, and press Update.

✅ Checkpoint: Lab NOC has two pages (the first plus Maps) and is shared read-only with Lab viewers.

7Open the host dashboard that came with the template

Go to Monitoring > Hosts and click Dashboards in the web01 row. You land on Linux Dashboard, which belongs to the template Linux by Zabbix agent active. Use the page tabs at the bottom to move between Overview and Problems.

The Overview page has gauges for CPU and memory utilization, graphs named CPU, Memory, Disk and Network, a Used storage pie chart and small value widgets such as uptime, load average, number of CPUs and the agent version. The Problems page holds a Problems widget and a Top triggers widget for this host only.

You didn’t build this; it appeared because the template carries it, and every host linked to that template gets the same page with its own data. You can’t edit it here, only in the template, which is what Lesson 10 does.

✅ Checkpoint: Monitoring > Hosts > web01 > Dashboards opens Linux Dashboard with two pages.

8Draw the “Lab network” map

Open Monitoring > Maps, click Create map, set Name Lab network, Width 800, Height 450, and press Add. Click Edit next to the new map to open the map editor.

Next to Map element click Add three times. Click each new icon and set it like this, pressing Apply each time:

Drag them into a line. Hold Ctrl, click zbx01 and web01, and next to Link click Add. Click the line between them, and under Link indicators click Add, choose the web01 trigger Linux: High CPU utilization and give it colour DD0000.

Set the link’s own Color (OK) to 00CC00, press Apply, and then Update at the top right to save the map.

Finally add a Map widget showing “Lab network” to the Maps page of Lab NOC and save the dashboard.

✅ Checkpoint: The map shows three elements and a green link between zbx01 and web01, and it sits on the Maps page of Lab NOC.

9Load web01 and watch everything react

The template trigger fires when CPU utilization stays above 90% for 5 minutes (min(...,5m)>{$CPU.UTIL.CRIT}), so you need about six minutes of load on web01’s single vCPU. Install stress-ng and run it for eight minutes:

Run this on web01:

sudo apt install -y stress-ng
stress-ng --cpu 1 --timeout 480s

You should see this after eight minutes (your PID will differ):

stress-ng: info:  [2214] setting to a 8 mins run per stressor
stress-ng: info:  [2214] dispatching hogs: 1 cpu
stress-ng: info:  [2214] skipped: 0
stress-ng: info:  [2214] passed: 1: cpu (1)
stress-ng: info:  [2214] failed: 0
stress-ng: info:  [2214] metrics untrustworthy: 0
stress-ng: info:  [2214] successful run completed in 8 mins

While it runs, keep Lab NOC open. Within a minute or two the honeycomb cell for web01 goes red and Top hosts puts web01 first; after about five to six minutes the problem “Linux: High CPU utilization” appears in Open problems and on the Maps page the web01 icon shows the problem and the link turns red. When stress-ng stops, the trigger recovers and the link goes back to green.

✅ Checkpoint: During the load the web01 honeycomb cell and the map link turned red, and both returned to normal after the run.