Community
- Command Central
- MVP Program
- Monthly Mission
- Blogs
- Groups
- Events
- Media Vault
Products
- Observability
- Network Management
- Application Management
- IT Security
- IT Service Management
- System Management
- Database Management
Content Exchange
- SolarWinds Platform
- Server & Application Monitor
- Database Performance Analyzer
- Server Configuration Monitor
- Network Performance Monitor
- Network Configuration Manager
- SQL Sentry
- Web Help Desk
Free Tools & Trials
Store

Polls Per Second Tuning - bad or good?

I installed the polls per second tuning application on my newly built 8.5.1 SP3 poller and it (a) broke the license and (b) caused me to re-run the config wizard (c) immediately started losing SNMP polls when I brought it up to the recommended levels of 99 & 61. I'm on a 3Ghz/6GB RAM system now with ooldes of GB's on a RAID 0+1, and the SQL server is separate. So to the question - with 5000 elements is polling at the default 30/sec ok and not using the recommended setting from the PPST application or should I expect to start missing polls to elements?

Find more posts tagged with

Accepted answers

All comments

jtimes

Orion itself is not CPU or Drive IO intensive, but it will hammer it's NIC, hopefully you have gig nics in it.

SNMP Collection intervals can not be set to lower than 1 minute, and with that many elements on one engine, it won't make it around the network if your chocking it through a 100MB

what are your Node Manager Settings for Polling and for Collection (Statistics)

I'd hazard a guess that the return SNMP bulkget (stats) packets are dieing on the server NIC buffer

What is your SNMP timeout set to in the Node Manager?

Is your DB server a monster as well?<---Thats where you need the CPU+RAM+Fast Disks and big NIC(s)

dave.woolterton

I would like to see more on these issues as they are clearly not highlighted in any of the Solarwinds documations I have seen over the years.

I have one poller with 8000 elements on it and colecting data every 5 mins. The DB Server is a stand alone box with a direct connection to the poller as well as the network. I did this for excactly the same reason to reduce the traffic flows on the NIC. All my NICs are 1GB and I still see data loss no matter what I set the polling to be.

FormerMember

Everything about the installation is default; collection times, timeouts etc. NTA is not being used, my GIG NICs are barely breaking 10Mb, Teamed with NFT'd between a pair of 6500's. SQL server is beefy too. I notice the problem when I review one of my 6500's and I don't see a bandwidth percentage next to an interface. Clicking the interface, I see an "Unknown or Unsupported Value" in the Percent Utilization resource and a fair amount of gaps in the charting.

From another support ticket, I was told that the Custom MIB Poller can sometimes take precedence over the NPM Statistics collection.

rajivbahl

Even i have a similar issue

The server resources are not highly utilized but the polls are getting missed or even the nearest connected router on same Gig LAN.

I have tried configuring the Polls per second to various value but the maximum i can achieve on polls completion is 99.14% i am currently using 50 50 for both the values. Recommended is 54 . I am planing to move to to 60 60 and check the change in polling. Require atleast one day polling to check the difference.

Regards

Rajiv