WMI statistic data inaccurate?

I had a similar issue and someone provided me with the following advice and it worked. If nothing else it can't hurt. 

I had some trouble getting WMI data from the scanners and needed some or all of this done:

dcomcnfg - pane Default Properties on my computer - Set default Impersonation level to Impersonate

In the folder: C:\WINDOWS\system32\wbem run these:

winmgmt /clearadap
winmgmt /kill
winmgmt /unregserver
winmgmt /regserver
winmgmt /resyncperf

wmiadap.exe /f

If it still does'nt work - stop WMI service - delete all content of folder C:\WINDOWS\system32\wbem\Repository and start service.

Hope you find this useful.

I have done all that, it still does not work.   Digging deeper, I get information from WIN32_PERFRAW, but not WIN32_PERFFormatted.  I found out that I can get good information from Formatted if I use a iRefresher.  I can verify that using wbemtest.

So my questions are these.  Does APM successfully pull information for anyone using any Win32PerfFormatted Enum and is anyone getting actual valid data using the Exchange 2007 template, the SQL Server 2005 (apm 2.0) template, or the IIS template?

If so, then I guess I have something wrong.  Else I wonder if APM is pulling good data.   All I get now are 0's.

 MarkS

I wanted to bump this because we are seeing the same thing.  At first we thought it was a rights issue, but we are able to get data using the same account and perfmon.  We are just starting to play with the APM and it seems cool but the fact that we're not getting data is not good.   We ran the "wmiadap.exe /f" command above because I found it in another post but that didn't seem to help.

I think I may know what is going on.  WMI does not work like snmp.   You cannot poll a 'cooked' wmi metric (win32_perfformatteddata) and expect to get anything worthwhile.   WMI was meant to be polled by a refresher that uses a time based value.  Meaning, If I poll formatted, it is expecting two things, the metric to pull, *and* the time frame to calculate.   So for instance, network traffic.   If you look at raw, you will see a counter that constantly rises.  But cooked, will be zero unless you tell it what time frame to calculate the delta.  So the traffic will come back zero unless you *tell* it I want this metric over 5 mins.

 I think that WMI monitoring is good, but there may be more involved to get what we want with what Solarwinds is giving us.   Really to me, Solarwinds should always pull raw, then graph based on time frame.   So when we set up a poller, say call it a delta poller, then have a configurable time period, then solarwinds always returns the delta or differenece in the raw between the two polls.   Graph and return that, and we are good to go.

Performance monitor has an (what MS calls) iRefresher in the code that is responsible for making good data out of this cooked data.  I am not sure that is what Solarwinds is doing.  

 This is after all my research.  I do not believe that some of the stuff that Solarwinds WMI poller is pulling is valid data.

Then, I could be off a mile :)

MarkS

 Salyerma,

 It looks like you may have found a bug.  Could you open a ticket so that we can investigate further?  If it's what we think it is, we may have a fix.

Will Do.  I will create one online. Case #64785

 MarkS.

Denny, I entered this case a couple of days ago.  I got one email telling me to use snmp instead.  I have not heard anything else.  Is there a hotfix to test?

Thanks,

MarkS

 I don't have an update yet.

Ok.  Thanks!

MarkS

Denny,  I am getting some push back from support.   It is their belief that there is nothing wrong with the software and that it is something on my end.  I am not sure that I am going to get much farther.  Do you understand what I am seeing or do I need to just let it go?  I may have mistakenly thought that you had an idea what might be causing my problem, but support is saying that nothing is wrong.

If it is something you are aware of, I will wait on a fix. 

Thanks,

Mark Salyer