Open mvp1108 opened 9 years ago
I have a same probloem, when use sum:1h-sum, the result is ok for me.
curl 'http://fe-perf08:8088/api/query?m=sum:1h-sum:counters.xxx.serverCount%7Bspanname=bbb,localhost=all%7D&start=2015/06/10'
[{"metric":"counters.xxx.serverCount","tags":{"source":"statsd","spanname":"bbb","localhost":"all"},"aggregateTags":[],"dps":{"1433876400":67044.0,"1433962800":68576.0,"1434049200":70467.0}}]
but when use sum:1d-sum, the first datapoint is missing, and <ts, value> is wrong matching
curl 'http://fe-perf08:8088/api/query?m=sum:1d-sum:counters.xxx.serverCount%7Bspanname=bbb,localhost=all%7D&start=2015/06/10'
[{"metric":"counters.xxx.serverCount","tags":{"source":"statsd","spanname":"bbb","localhost":"all"},"aggregateTags":[],"dps":{"1433894400":68576.0,"1433980800":70467.0}}]
Pull request #657 should address this problem. Just waiting on @manolama to acknowledge and merge it. We have done some significant functional testing, and it seems to work fine.
When downsampling to "daily", the timestamped values always bear a timestamp of UTC midnight, but I am interested in seeing values like sum and average for the calendar day as defined in the local timezone. I have not been able to figure out a way to do this -- is there one?