Preamble

The problem reported through this ticket is very similar to the ZBX-12821 problem which has been fixed more than a year ago.

Context

Zabbix version 4.0.15. Google Chrome version 79.0.3945.117. Firefox version 68.3.0.

A Zabbix server monitors around 40 hosts. Around 30 of these hosts are shut down which is producing around 30 "Unavailable by ICMP Ping" problems.

A dashboard is configured and composed of: a map widget, a map navigation tree widget, a problems widget and 2 graph widgets.

The frond-end Zabbix web interface can be reached from the hosts monitored by the Zabbix server.

Steps to reproduce:

1. From a first host: open the Google Chrome web browser, access to the Zabbix web interface, let 1 tab over the dashboard and open 5 tabs over the Monitoring/Problems page.
2. Let it go for couple of hours (in this case around 18 hours)
3. From a second host: open the Firefox web browser, access to the Zabbix web interface, let 1 tab over the dashboard and open 5 tabs over the Monitoring/Problems page.
4. Let it go for couple of hours (in this case around 18 hours)

Result:

For both hosts, the memory utilization increases till reaching very high values (see attached screenshot). The memory utilization is calculated for each host by the Zabbix server comparing UCD-SNMP-MIB:memAvailReal and UCD-SNMP-MIB:memTotalReal.

The memory utilization can also be checked using the top command on both hosts (see attached screenshots). We observe that chrome and firefox/Web Content processes generate most of the memory utilization.

The memory can be released by closing the web browsers on each host.

Expected:
No memory leak.