Question regarding organization and load balancing.

lcsw2013

Hi,

I'm trying to figure out a way to better organizing our SolarWinds. I've made a feature request I'm trying to move some momentum behind for an auto load balance feature. For example, in discoveries. If you don't actively select the poller to which you want to discover too, all discoveries go to the primary as it is the default. If we have a new guy not familiar in solarwinds, or someone else assisting (Which happened to us), you end up with an overwhelmed primary and additional pollers that have no devices assign to them. If we have an option to choose between manual and auto with auto being default. The system could run a process similar to database maintenance where it decides based on load numbers which device would go where. This way you never have to worry about a overwhelmed poller again.

But I'm here to ask anyone reading this how do you organize your solarwinds or load balance it across the pollers with out so much manual work? Is there a way to accomplish this? If the feature gets considered that would be ashmed and it would end this question but since the feature isn't developed yet I need to ask and get ideas and opinions.

I'm drowning in master sheets and spreadsheets. And the numbers are more likely to get skewed due to the number of spreadsheets I have to swim through. Then upper management is on neck about variances in numbers and what can we do to always have accurate counts and readings of solarwinds health.

Secondly, I'm curious to know if there is a way to develop a report that would give me an across count of devices in ICMP mode. There is always a difference between unknown count and icmp count even though technically they are unknown and not snmp devices. Then come to find there are snmp devices and icmp devices that aren't unknown but are in the icmp category. This is the point I start seeing doubles and have to give up trying to get an accurate report. I've created and edited several of them. But none give me the expected output. And the variance in numbers is way to big. Any help here would be great.

Lastly, if I have a switch for example that is part of a name change project, but the ip stays the same, will solarwinds automatically scan and change the name accordingly or does it stay the same. I lean towards there being a change but am not sure.

Appreciate any help in advanced. Thank you!

Find more posts tagged with

Reports

load balance

network performance monitor

organization

npm

Accepted answers

All comments

mesverrum

I usually build myself a home view called SW Admin and have a collection of reports I've built up over the years that track these kinds of things for me. I'll upload some of them tomorrow but the first thing on the page is a chart showing the polling loads on each server so I can watch for balancing issues, I've never worked in an environment that grew so quickly I couldn't spot a problem coming just by watching that chart. If things don't change too much then manually balancing once every few months is not a big problem. If things are very dynamic you can write a sql script with logic to decide who to move and under what conditions. Such as leaving network devices on their current poller because we haven't added the ape to an acl yet.

The report of devices in icmp is simple, create a node report and filter the objectsubtype = icmp, that field is basically the polling method.

Your caption will not update automatically if you rename the device. It gets set automatically when you add new devices but the sysname will update when the next rediscovery happens. The caption only sets itself once so that it doesn't erase changes you may have made.

How Caption field for nodes is determined - SolarWinds Worldwide, LLC. Help and Support

I typically have reports showing cases where sysname/caption/hostname don't match up the way I'm expecting so I can keep an eye out for those kind of changes.

lcsw2013

Thank you! Great points. Especially the admin view page. If you don't mind me asking, how did you setup the charts for visuals of the polling loads? Any screenshots you can share by any any chance? And any basic script you might wouldn't mind sharing? This will help me kick start something similar to help me monitor.

One issue I keep having is ephemeral port exhaustion on some servers I can't seem to fix. Would you happen to by any chance have any tricks for this? I'm almost convinced it could be the network card setup on the vm's for our solarwinds monitoring machine. And if so I want to make changes so the server can keep up with the infrastructure we have to monitor.

Again, any help is more than appreciated. Thanks!

mesverrum

This is a custom chart I use for polling loads that uses a custom SQL query as its datasource

select
  --This is the Time Column for both of your data series in your custom chart
  d.[Date]
  --This will be the 'ObjectID' in your custom chart and the 'Caption Column' in your custom chart
  ,e.servername as [Server]
  --NPM Polling Rate from Subquery 'x' below
  ,avg(x.avgNPM) as 'NPM Polling Rate'
  --SAM Polling Rate from Subquery 'y' below (comment out/remove if SAM is not installed)
  ,avg(y.avgSAM) as 'SAM Polling Rate'

from pollercapacity_daily d
join
--This section is used to get the NPM Polling Rate
  (
  select
  [date]
  ,engineid
  ,avg(scalefactor) as 'avgNPM'
  from pollercapacity_daily
  where throttlinggroup = 'Orion.Standard.Polling'
  and [date] > (getdate()-30)
  group by [date], engineid
  ) x on x.engineid = d.engineid and x.[date] = d.[date]
--This section is used to get the SAM Polling Rate (comment out/remove if SAM is not installed)
left join
  (
  select
  [date]
  ,engineid
  ,avg(scalefactor) as 'avgSAM'
  from pollercapacity_daily
  where throttlinggroup = 'APM.Components.Polling'
  and [date] > (getdate()-30)
  group by [date], engineid
  ) y on y.engineid = d.engineid and y.[date] = d.[date]
join engines e on e.engineid=d.engineid


group by d.[date], e.servername, d.engineid
order by d.[date] desc

This is a SWQL query that I use with the Custom Query resource that shows me how my pollers are all doing

SELECT ServerName, IP, ServerType, Elements, Nodes, Interfaces, Volumes, PollingCompletion, MinutesSinceKeepAlive,
CASE
WHEN PollingCompletion < 98 THEN '/Orion/images/ActiveAlerts/Serious.png'
WHEN PollingCompletion > 98 THEN '/Orion/images/ActiveAlerts/Check.png'
END AS [_IconFor_PollingCompletion]
, CASE
WHEN MinutesSinceKeepAlive > 2 THEN '/Orion/images/ActiveAlerts/Serious.png'
WHEN MinutesSinceKeepAlive <2 THEN '/Orion/images/ActiveAlerts/Check.png'
END AS [_IconFor_MinutesSinceKeepAlive]
,CASE
WHEN Elements> 11500 THEN '/Orion/images/StatusIcons/Small-Down.gif'
WHEN Elements< 11500 and Elements > 9000 THEN '/Orion/images/StatusIcons/Small-Warning.gif'
WHEN Elements<  9000 THEN '/Orion/images/StatusIcons/Small-Up.gif'
END AS [_IconFor_Elements]
,'/Orion/Admin/Details/Engines.aspx' as [_linkfor_ServerName]
FROM Orion.Engines


order by ServerName

-Marc Netterfield

Loop1 Systems: SolarWinds Training and Professional Services

LinkedIN: Loop1 Systems
Facebook: Loop1 Systems
Twitter: @Loop1Systems

mikesky

Sorry to bump an ancient thread. I got the table setup successfully. When setting up the chart, I get an error. If I run the code in SWQL Studio I get:

"mismatched input 'NPM Polling Rate' expecting 'Identifier' in Select clause"

I'm not that familiar with SWQL. Any thoughts? Thanks