Capacity Metrics App - Inconsistency when using drillthrough to see timepoint of interactive Cu

[dbeavon3]

For a period of time (yesterday and today) whenever I tried to drillthrough to the timepoint details for a 5 PM spike, I was not being shown the complete analysis for that point of time

We are supposed to be able to analyze these spikes by right-clicking and drilling through to the "timepoint details".

However it wasn't working as expected.   The background operations would appear but the grid showing "interactive operations" would remain blank (no wait cursor either).   No matter how many times as I tried, it wouldn't work.  I waited hour after hour, thinking there was a backlog and it wouldn't work.

Is there an SLA that would determine how long to wait for the timepoint details (interactive operations)?  Is there another place to get that information in a more reliable way, preferrably without the use of the capacity metrics app?  Do people rely on XMLA profiler traces, in order to gather their own history of the interactive operations?

Eventually -  as a hail mary -  I clicked the reset filters button at the top of he app, and things started working again in the way I remember. 

Has anyone else seen this sort of unpredictable behavior when drilling into timepoint detail?  Perhapse there was a filter in effect that I wasn't aware of?  Why was there an inconsistency where I was getting incomplete results (the missing timepoints only impacting my "interactive" operations and not the "background" operations)?  Any feedback would be appreciated.  I've never been a huge fan of this metrics app.  Considering how much customers are paying for Power BI, it seems that Microsoft has an extremely low bar when it comes to building management tools for the service. 

[dbeavon3]

Hi @lbendlin 

Thanks for the reply.   I will try to investigate the drillthrough in more detail next time.  I just think it is weird that some filtering was in effect, without a means to be aware of it.  I'm guessing that I'm missing a best practice for using PBI apps, like maybe I should be resetting my filters once a week or something.

I really wish there was an API for getting the timepoint information. 

... The only alternative I can think of is using XMLA-based tracing, but that takes a bit of effort.  I noticed that even SQL profiler has a hard time running traces against PBI premium for more than a few minutes at a time I suspect Microsoft is discouraging customers from doing that.  In any case it would probably just take away from our limited CU's.

I found the version and it doesn't say v17.

Microsoft Fabric Capacity Metrics app
Help tab ->Release Notes
Says version 1.2.1

[lbendlin]

Help tab? What Help tab?  Are we talking about the same thing?

Microsoft Fabric Capacity Metrics

[dbeavon3]

Hi @lbendlin 
This is the version.  The number is not even close to v17.  

[lbendlin]

The latest is 1.5.1  which is v21.  Yours feels more like v14 🙂

[dbeavon3]

Thanks.  I really wish they would create an admin console that can be used from the desktop client (like PBI desktop).  Keeping this console up to date is more trouble than it should be...

[lbendlin]

I wish they gave us direct access to Kusto so we can create our own monitoring app.  I wish they could make it more real time and even proactive (like the Monitoring Hub).  As it is now it is reporting old news (on average five to eight minutes late) AFTER the brown stuff hit the spinny thing.

[lbendlin]

v17 ?  Be aware that there is a "TimePoint Detail"  drillthrough page, AND an "TimePoint-Detail" drillthrough page (!!!) .  This whole app is a disgrace, half baked, and obviously frankensteined together from incoherent pieces. Just look at the data model.  Feels like two versions are fighting for attention.

Since you have a Pro license you should open a Pro ticket at https://admin.powerplatform.microsoft.com/newsupportticket/powerbi

The underlying storage is Kusto based, and not accessible to us.