SolarWinds agent status becomes unknown about 2 or 3 times a week

Question

Hello,

We recently changed from WMI to agent polling in our environment. We currently poll about half of our devices trough agent (1179 nodes) both linux and windows, varying from Centos, Redhat and Ubuntu to Windows 2003 and above(2016 included)

About 2 or 3 times a week about 500 agents become unknown and stop polling data for nodes

A restart of the agent solves this issue temporarily, but we are yet to find a root cause or a fix for this.

This are the module versions that are currently in use in our environment:

Agent 1.7.0.331

NPM 12.1.5300.0

SAM 6.4.1020.0

We have a main poller and 3 additional ones.

We have parsed trough the agent logs on machines with nothing that would point to an error or a failure.

Has anyone else experience similar issues? And if so, do you have a root cause or a fix for this type of problems?

tech2734 · Answer

Check the permissions of the folder /opt/SolarWinds/Agent/bin/appdata/Logs - we found many of our application templates had unknown status until a restart (temporarily), but changing the folder permissions for world r/w seemed to fix it permanently.

cahunt · Answer

Poll using WMI / No Agent. If Access is setup properly then you will get the same results.

Others  have opened support cases to get help determining what the issues with the agent are within each environment.

hpstech · Answer

Windows Agents are unreliable for us as well.

Its been an epic challenge keeping them green.