Turning on the performance monitor in PAW

[Steve Rowe]

Hi,
I don't have any of the legacy clients available to me and so have no way to see the cube sizes and also I cannot see an option to turn on the performance monitor so can monitor and track cube sizes.

Anyone seen an option to enable this anywhere?

Thanks in hope!

[Elessar]

Hi

Generally speaking you can, but you need to go roundabout ways: in Administration, server configuration, you can change some tm1s.cfg parameters including PerformanceMonitorOn. PerformanceMonitorOn is dynamic, so this will turn it on immediately.

[Steve Rowe]

You can turn on the option in the cfg editor of PAW in Administration / YourDB / Configuration / Default / PerformanceMonitorOn

Unfortunately this method requires a restart of the DB if you change it through the console.

Although the documentation states if you change it in the cfg it is dynamic. (I don't have access to this though...)

Restarting will destroy the very information I am trying to analyse of course, so super-impractical....

Aren't Friday afternoons great!

[Elessar]

I've checked this now: it does turn perfmon On without server restart (literally PAW adds "PerformanceMonitorOn=T" row in tm1s.cfg file)

[Steve Rowe]

Yeah, did I mention it was Friday afternoon??

Never trust the documentation!

Now I just need to understand that fact that

Cube RAM totals 1.5GB
Server RAM =10.5GB
Garbage = 0.8GB
Taskmanager states 10.5 GB

So which is wrong...Garbage or Cube.....

Thanks for your help Elessar, have a great weekend!

[Wim Gielis]

You could also use the REST API to change settings.

[gtonkin]

... So which is wrong...Garbage or Cube.....

If there is a process running, you would not see how much memory it is consuming with transactions in case it needs to roll back.
And then don't forget that all the dims, sets etc. eat memory too - can still use the properties window in architect from the rich environment if you need to check dimension memory usage.

Failing that, could be a memory leak, there apparently were some issues fixed in PA 2.0.9.17 and we have recently upgraded our cloud Dev environment to check this out as we have had similar issues doing the math and coming up short, even on afternoons other than Friday.