Incorrect datastore graph stacking with v11a

[DonZoomik]

After v11a upgrade datastore graphs look broken. The best I can describe it, is that all stacked lines are of roughly equal thickness and graph scale is incorrect, sometimes orders of magnitude.
Example image with stack by host shows it best:

Now the same context (all vCenter datastores) in VM stacking view (few minutes offset):

Apparently we've gained a few million IOPS by changing stacking type.

Real IOPS is 15-30k.

I've seen it in 2 environments now but havent bothered to create ticket until today when I could have really used the graph to analyze some IO stuff...
Anyone else?
#05143924 (Community edition)

[HannesK]

Hello,
I don't know how it was before 11a, but I think it looks the same for me.

This is "stacked by hosts" and and I see something around 3000 IOPS in the graph



this one is "stacked by VMs" and I see something around 120k IOPS in the graph



I will check with QA and come back on this.

Best regards,
Hannes

[DonZoomik]

With v11 non-a and before it used to show sane numbers.
As with your datastore stacking, small number of data rows seems to be more-less sane but VM stacking is crazy. You should probably see a few thick lines for few VM-s with stable high IO and a lot of thin ones for VM-s with dimishingly small IO with peaks for maximums. Instead you have indistinguishably equal small lines while graph scale is going insane. In both views, graph scales should be rougly equal as you're just slicing the same data differently.

[wishr]

Hi Don,

Our QA team is looking into it.

Update: We've identified several issues with these graphs and will apply the necessary changes in future VONE versions. Unfortunately, these issues require a complex approach to be fixed, so creating a hotfix is not possible.

Thanks