Main Polling Engine - Best Practices

I would like to get best practices for the main polling engine.

Currently we are having some issues with the CPU usage and polling stability that we have been working with Support on for months since upgrading to 2024.2

However, we do have all of our orion servers (include additonal polling engines and webservers) added to the main polling engine for polling.
How do others handle this? We do not have HA and our platform is on premise.

Parents
  • 1. I am sure you would have seen this https://documentation.solarwinds.com/en/success_center/orionplatform/content/core-optimization-polling-engines.htm

    2. As you are already aware of, MPE is all in one be it your Web Server, Alerting, Polling, Reporting etc - hence with respect to polling if possible keep the number of devices as less as possible like say half of what they say around 6000 elements.

    3. Since you have mentioned you have AWS - thats good to know as you lift of the load for user login (Web Server/IIS load off).

    4. Make sure you always patch your SolarWinds MPE on time and reboot it atleast once in a month.

    5. As well keep track on disk space (clear temp once in a month, make sure you have sufficient disk space) and any sort of Antivirus on your MPE - try excluding SolarWinds folders from the scans or Defender or local firewall etc

    6. If you have any external integrations that hits your platform through SolarWinds API's try avoiding too many calls on MPE.

    7. Lastly, dont run additional softwares on MPE and make sure too many people dont login into MPE through RDP and if they do its best to log them out on a daily basis - If possible make sure only few RDP connections are allowed and admin list is small.

    8. Under 'My deployments' check for issues if any on MPE.

    9. Always make sure latency is low between your MPE and DB/Others platform components.

    10. If at all you can, check if network latency is fine from MPE to the end devices that you monitor

    11. Other things like dont create complex reports and even though you do restrict access to admins only through limitations, if you create reports which produces massive reports and if all of your users have access to it that might choke your instance sometimes, as well check the amount of UNDP's or unmanage schedules on MPE etc etc - Anything and everything that you are using your MPE for :) and yes Netpath if you are using that check how many of it exists. There are a few more things that I can think of (like number of syslogs, traps etc)  but ...  this should be a good starting point for you... 

    Hope this helps.

Reply Children
No Data