-
Problem report
-
Resolution: Unresolved
-
Trivial
-
None
-
5.0.21
-
None
-
Debian 10 VM
2vCPU, 4GB RAM
MariaDB
-
0.125
Issue has been discussed here:
Sometimes change per second calculation produces very big values (terabits per sec on 1gigabit interface). Usually faulty values are related to outbound traffic (ifhcoutoctets, ifoutunicastpkts). Raw snmp counter values seems to grow steadily, snmpwalk shows also reasonable values.
In my case faulty values didn't show up after bulk requests was disabled. Another user says that disabling bulk requests made faulty values more rare.
Could there be some overflow etc. which could mess values with bulk requests?
In my case there is Cisco IE-5000 switch with ~700 items. Polling cycle of items is 5min. Zabbix server has less than 100 NVPS.
Problems can be seen with following Zabbix versions
4.0.16, 5.0.10, 5.0.19, 5.0.21.