1 machine constantly "discovers" different cpu

Hi,
over last few months all issues I used to have with my librenms on rpi4+ssd setup sorted out and it was happily humming along. However few days back I spinned up a new linux VM (LXLE based on ubuntu 18.04), added it to librenms, and this single machine keeps swapping its processor data from correct (4x xeon) to 1x ucd-old cpu (w/e that is). This causes massive gaps in graphs obviously as I only get graph for when it thought there were 4 cpus or only a graph filling the gaps where it thought it was just 1 cpu.

I can force rediscovery which usually switches in the correct processor, but after half a day or so the cpu is usually swapped.

Screenshot here:

Validate output:
librenms@nms:~$ ./validate.php
====================================
Component | Version
--------- | -------
LibreNMS  | 1.60-92-g7254f13b2
DB Schema | 2020_02_10_223323_create_alert_location_map_table (159)
PHP       | 7.2.9-1+b2
MySQL     | 10.3.22-MariaDB-0+deb10u1
RRDTool   | 1.7.1
SNMP      | NET-SNMP 5.7.3
====================================

[OK]    Composer Version: 1.9.3
[OK]    Dependencies up-to-date.
[OK]    Database connection successful
[OK]    Database schema correct

I don’t know how to pick single data point from snmp so I posted the snmpwalk output on my cloud:
https://turbomrak.ddns.net/s/5qyoBoAkJRi7QfY

Any help appreciated, thanks!

OK, this has stopped as of last few days (no idea in relation to what as I’ve done nothing), however now another machine (raspberry pi 3b+) running pi-hole does this once or twice a day:

|2020-03-10 23:02:07|system|[pi-hole]|Device Icon changed linux.svg => raspbian.svg
|2020-03-10 23:01:14|system|[pi-hole]|Device Icon changed raspbian.svg => linux.svg
|2020-03-10 23:01:14|system|[pi-hole]|OS Features -> Raspbian GNU/Linux 10
|2020-03-10 23:00:19|system|[pi-hole]|OS Features ->

I have 2 identical rpi 3b+'s runnig pi-hole, only one of them does this.

I use 1min polling, but at no point around when this happens or at all do the pollers not poll a device within a minute, all devices are finished polling in under 30sec.

This may happen if snmp replies are coming slow or erraticaly. Can you confirm the connectivity is always good ?

The connectivity should be allright, all the other extends (raspberry, osupdate, entropy, pihole, ntp, apcupsd) work fine with no gaps in graphs. The pihole on which it happens is my primary dns + ntp server, average load on it is under 0.2.

The pollers never take longer than 30ish seconds to finish. I’ve tried increasing the polling interval nonetheless to no avail.

this is the graph of polling interval showing its finishing on time:
https://turbomrak.ddns.net/s/scAqSj6nGgTiinM

I’ve no idea what else to check, but I think if there were any connectivity problem it’d affect my internet but it doesn’t.

In device / edit / snmp there is 2 text input boxes saying seconds and retries I take it these are timeout and retries for the poll, I typed 29sec and 2 retries there and it hasn’t happened last 24h so far. So it looks like there’s too many extends and the snmp output gets cut off or something to that effect, although none of the monitored extends has gaps.

edit: 3 days later it hasn’t happened since so i’m marking this as answer