There are gaps for several seconds or even minutes in data that are collected by SNMP.
In some cases, all data is lost (for all of monitored OIDs).
But in others, only a few of them.
Some information about environment:
- Mmosa Firmware - 1.4.6
- SNMP v2c
- net-snmp-libs.x86_64 1:5.7.2-24.el7_3.2 (CentOS 7)
- NMS Zabbix 3.2.0
- Amount of monitored OIDs - 58 (per single mimosa), including:
- updated every 60 seconds - 50
- … every 3600 secs - 1
- … 86400 seconds - 7
As far as I understand, this is not higt loaded (by SNMP) instances. And it seems that problem not on the NMS side. I tried to run in console some looped queries like this:
$ while true; do date; snmpget -v2c -c <FOOBAR> <MIMOSA IP> <SOME OIDS HERE>; sleep 5; done
and got the same results - some times I got just
Timeout: No Response from <MIMOSA IP>
We observed this behaveour on at least 4 mimosas (of 4 tested) in our batch. Also I noted that such gaps happen more often if I increase the frequency of queries.
Can you please explain what we can do to perform SNMP monitoring in correct and reliable way?