https://community.netapp.com/t5/Software-Development-Kit-SDK-and-API-Discussions/perf-object-get-instances-counter-rollovers/m-p/30831#M764 <P>Could the API documentation for perf-object-get-instances cover the proper handling of counter rollovers?</P> <P>&nbsp;</P> <P>Some high-rate-of-change counters (e.g. system:system:disk_data_read, system:system:disk_data_write) are highly susceptible to rollover, and naive implementation of the SDK will result in some negative rates. At least for those, a negative value is nonsense, but I don't see where this assumption could be generalized.</P> <P>&nbsp;</P> <P>Would it be correct that counters are always positively incrementing, in which case "Raw(T0) &gt; Raw(T1)" would be sufficient? If not, is there a subset of counters (e.g. properties "rate" or "average") that could be expected to always be positive?</P> Thu, 05 Jun 2025 06:02:55 GMT kevingraham 2025-06-05T06:02:55Z https://community.netapp.com/t5/Software-Development-Kit-SDK-and-API-Discussions/perf-object-get-instances-counter-rollovers/m-p/30831#M764 <P>Could the API documentation for perf-object-get-instances cover the proper handling of counter rollovers?</P> <P>&nbsp;</P> <P>Some high-rate-of-change counters (e.g. system:system:disk_data_read, system:system:disk_data_write) are highly susceptible to rollover, and naive implementation of the SDK will result in some negative rates. At least for those, a negative value is nonsense, but I don't see where this assumption could be generalized.</P> <P>&nbsp;</P> <P>Would it be correct that counters are always positively incrementing, in which case "Raw(T0) &gt; Raw(T1)" would be sufficient? If not, is there a subset of counters (e.g. properties "rate" or "average") that could be expected to always be positive?</P> Thu, 05 Jun 2025 06:02:55 GMT https://community.netapp.com/t5/Software-Development-Kit-SDK-and-API-Discussions/perf-object-get-instances-counter-rollovers/m-p/30831#M764 kevingraham 2025-06-05T06:02:55Z