In a 4-node Couchbase CE 7.0.1 cluster, the Total Scan Rate at indexes only goes up, as if it would be some sort of an accumulate. However, documentation and the UI obviously states that it is a “per second”-metric.
When I look at the Scan Rates on individual indexes, the graph on the UI there shows changing Scan Rate “/seconds” values, which are much lower than 6 million.
@zoltan.zvara this should be derived from the avgScanRate (avg_scan_rate) statistic. Looking at this in Index code I saw it does have a calculation problem for partitioned indexes (not correctly computing the average across partitions) which I have added to the existing defect for a couple other things that had the same problem: MB-44241. However, this does not look like it would cause such an error as you are seeing.
There could be a problem in the UI’s rendering of this stat. From the symptoms this sounds more likely to me. Hopefully @eben can lend some UI expertise.
I’m not the expert on stats, but this does appear to be a bug where it’s summing values instead of showing rate. @pavel can say for sure.
It looks like a bug since it should sum values and show rate. We will fix this in 7.1.
This is also a bug in EE and for many other values that should be showing rates.
