Is there any documentation for these metrics?

We're trying to troubleshoot some analysis related issues and would like to better understand the metrics available.

In particular, I'm trying to figure out what the trailing number is on several metrics:

For example: NumAnalysesInExeQueue_0 , NumAnalysesInExeQueue_10 .... 50 .... 100 ... 500 etc.

I'd appreciate any help

Thanks!

FromTemplate | 2223 |

WithNoTemplate | 3 |

WithOverriddenTimeRule | 0 |

PerformanceEquation | 2213 |

EventFrame | 0 |

Rollup | 13 |

OtherAnalysisRule | 0 |

Periodic | 129 |

Natural | 2097 |

OtherTimeRule | 0 |

RunningStatus | 2182 |

ErrorStatus | 44 |

OtherStatus | 0 |

Rank_0 | 2182 |

Rank_1 | 0 |

Rank_2 | 0 |

Rank_3 | 0 |

Rank_4 | 0 |

Rank_8 | 0 |

Rank_Max | 0 |

FractionCPU | 0.0909096736596737 |

PrivateBytes | 2263293952 |

VirtualBytes | 18803884032 |

EvaluationThreads | 4 |

EvaluationQueueCount | 0 |

InputsCached | 8148 |

OutputsCached | 1708 |

EventsProcessed | 378141 |

EventsCached | 236483 |

CacheHitCount | 247268 |

CacheMissCount | 240 |

EvaluationCount | 50688 |

SkippedCount | 0 |

EvaluationErrorCount | 6851 |

TotalEvaluationCount | 57539 |

AFValuesWritten | 12777 |

EventFramesCheckedIn | 0 |

Latency_0 | 0 |

Latency_5 | 108 |

Latency_10 | 0 |

Latency_20 | 0 |

Latency_40 | 0 |

Latency_80 | 1708 |

Latency_Max | 418 |

NumAnalysesInExeQueue_0 | 5 |

NumAnalysesInExeQueue_10 | 0 |

NumAnalysesInExeQueue_50 | 0 |

NumAnalysesInExeQueue_100 | 0 |

NumAnalysesInExeQueue_500 | 0 |

NumAnalysesInExeQueue_1000 | 0 |

NumAnalysesInExeQueue_5000 | 0 |

NumAnalysesInExeQueue_Max | 0 |

TotalNumOfOutputsNotSavingHistory | 920 |

TotalNumOfOutputsSavingHistory | 1729 |

TotalNumOfOutputsNotSavingHistoryUsedAsInputs | 347 |

NumOfAnalysesWithOnlyOutputsNotSavingHistory | 468 |

NumOfAnalysesWithOutputsSavingHistory | 1758 |

NumOfAnalysesWithOnlyOutputsNotSavingHistoryUsedAsInputs | 0 |

NumOfAnalysesWithOutputsSavingHistoryUsedAsInputs | 2226 |

NumAnalysesWithTriggerTimeOutput | 2226 |

NumAnalysesWithExecutionTimeOutput | 0 |

NumAnalysesWithFixedPositiveOffsetOutput | 0 |

NumAnalysesWithFixedNegativeOffsetOutput | 0 |

NumAnalysesWithVariablePastOffsetOutput | 0 |

NumAnalysesWithValidOtherTimeOutput | 0 |

Those are the summary performance counters for the Calculation Groups that are created by the PI Analysis service.

Each calculation group keeps a track of what it last completed, one of those is the number of analyses in the execution queue.

You can use those statistics to see if your analyses are all bunched together (calculated together).

I forget how the calculation grouping is performed, but I think it was by time rule scheduling.

So to explain...

When you "Refresh" those statistics each of the Calculation Groups (which you can see in "Service details => EvaluationStatistics => EvaluationStatisticsForCalculationGroups") is checked for the "NumberOfAnalysesInExecutionQueue", then it increments the appropriate counter. So if the first group reports 0 analyses in execution queue, then the "NumAnalysesInExeQueue_0" is incremented. If it reports "6" then the "NumAnalysesInExeQueue_10" is incremented, and so on. Anything over 5000 gets dumped in the "_Max" counter.

If the majority of your analyses execute together then you'll probably see your max counter jumping up and down...but obviously depends on the number of analyses configured in the first place, and the scheduling.

That was the unofficial explanation anyway. Hopefully Mike Zboray can give an official description.