Join GitHub today
GitHub is home to over 20 million developers working together to host and review code, manage projects, and build software together.
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
Already on GitHub? Sign in to your account
add custom metric types #23
Conversation
|
@suyash looks OK, just one issue. In PCP counter semantics, and in system performance analysis in general, counter (MMV_SEM_COUNTER) values never go backwards. I'd recommend dropping all decrement operations, because the PCP client tools will interpret this as an error (counter wrap) and report no value for that sample, which will be very confusing to the user. Any value that can go backwards should be modeled as a gauge - in PCP, these map to instantaneous metrics (MMV_SEM_INSTANT), and then the client tools know what to do with 'em. |
suyash commentedAug 3, 2016
•
edited
Potential types to consider
DeriveRate (Meter)RateandDeriveare supported implicitly by PCP, just set the reporting rate at the time unit (seconds, minutes, hours...) and set semantics toInstantSemanticsto get Rate andCounterSemanticsto get Derived.Gaugeis a shorthand to get afloat64metric withInstantSemantics, whileCounterimplements auint64metric withCounterSemantics.HistogramSummaryto be considered after #28
Reservoircouldn't find a real world example of applying reservoir sampling on metric values, why add a value if the probability of it having any real impact just keeps diminishing
see: