SolarWind polling Engine collector service consumes excessive CPU approx. 50% average. the polling engine is having 8 Cores & 36 GB RAM and polling approx. 20000 elements. Unable to find the issue of HIGH USAGE.
20k elements is HIGH
The collector service is the one that does stuff, if that's consuming 4 cores for 20000 things it's probably doing a pretty efficient job
What solution I have because it is consuming 100% CPU usage through overall process.
You want the elements per poller closer to 10-12k really, could get a APE
CPU cores should probably be closer to 12, could upgrade the box
Could go much higher in theory but I dont recommend it.
You could call support if it's spiked and nothing's changed
Could be virtualization issues
If everything works fine and it's not unusual and nothing's bugged you could decrease polling frequency, or try and make the polling easier
Install Additional Polling Engines, if your licensing allows it, as Adam mentioned 10-12k elements on the MPE would be better on performance.https://documentation.solarwinds.com/en/success_center/orionplatform/content/install-scalability-engine.htmhttps://documentation.solarwinds.com/en/success_center/orionplatform/content/orion_platform_scalability_engine_guidelines.htm
Actually I am facing this issue in one of additional polling engines. I do have some other additional polling engines having same configuration and elements but there the resource utilization is normal like say average 50-60% CPU. Any specific thing that need to be taken care on the questioned polling engine ?
This might help -
Repair a failing Job Engine, Collector, and Information Service in the Orion Platform (site.com)
But I would still suggest you connect with SolarWinds and raise a case with them.
__PRESENT__PRESENT__PRESENT__PRESENT
SW recommends keeping the number of elements per poller under 12,000, more than that they recommend an additional polling engine.
Additionally, check the polling frequency as it may affect both collector and jobengine. Consider slightly decreasing the polling frequency in the Polling settings. Configure polling interval settings in the SolarWinds Platform
You can also check the collector logs for any abnormalities, such as too many failed requests to the same node or interface: C:\ProgramData\SolarWinds\Collector\Logs.
But as per solarwind article I can poll that many elements correct ?
Addionally I Do have other polling engines with same configuration but those are polling much more elements with out any issue?
Anything specific with the problematic polling engine ! There is case opened since more than a month but no resolution yet.
there could be several reasons to it, its not just the polling element count like 12K limit.
1. Depends on where your pollers are placed.
2. What is the latency between your pollers and devices.
3. What route does it take to reach the end device.
4. If you have more numbers of devices that are down in that poller or say suppose you have more of devices that are currently having communication issues like SNMP, WMI, ICMP etc it adds to the load and impacts your poller performance.
5. Then there could also be a chance that certain devices are taking more time than usual to respond back to SolarWinds polls.
6. Then there could be issues like temp files not cleared on your SolarWinds Servers or windows patching is pending on your SolarWinds Server or When did you last reboot your SolarWinds Server or have you ever done the recommendation that I provided in my previous post ? etc etc
7. There could be several others reasons as well which might contribute to performance issues, if you know your setup and environment well including the geographical location of your devices plus your boxes plus network connectivity plus your SolarWinds servers you would be able to crack this. Always remember there are 'n' number of factors that can cause performance issues on your box.
Hope this helps.
Thanks vinay. Actually I did repair the SolarWind Platform application as per solarwind recommendation but result negative. Let me look after the points you shared. i will posted further.
It's normally 12k per Polling Engine. We have to consider other things as well, eg. machine resources, polling frequency, number of agents assigned to this server, SAM applications, VMs, and so on. As Vinay mentioned, there are several reasons for the collector to use that amount of CPU. I once saw a very small thing causing issues to collector and job engine (CPU usage): the default logging level was set to debug. It's worth checking it.Also, if you don't get it resolved soon you can try to escalate the case to a higher level in support.
Potential Causes and Solutions for High CPU Usage in SolarWinds Polling Engine:
Polling Rate: Check the polling intervals. Reducing the frequency can help lower CPU usage.
Resource Allocation: Ensure adequate resources are allocated to other processes.
Database Optimization: Look into database performance. High CPU usage may be related to slow database queries.
Element Distribution: Consider distributing elements across multiple polling engines to balance the load.
Updates: Ensure that all software and hardware drivers are updated, as performance issues can arise from outdated versions.
Diagnostics: Use SolarWinds diagnostics to identify specific processes consuming excessive resources. KW glasballon deko ideen
I'm going to go ahead and remind everyone in this thread that anyone on the HCO license gets quad stacked pollers by default. So, assuming you give the server enough CPU and RAM you can now poll 48000 elements per polling engine but it needs a lot of horsepower to actually do it. In my old environment my quad stacked APE's all had at least 12 CPU and 24 GB RAM, but the specific amounts required depend on what kind of polling you do. 8 CPU is not going to be enough for 20k elements though unless you have a very slow polling interval, so you just need more CPU flat out.documentation.solarwinds.com/.../orion_platform_scalability_engine_guidelines.htm
