[FIX] thold 0.4.3-2

[deu439356]

Thanks for the information.

the longer ago the reference value -> the lower the chance you get a false positive

Any suggestions to get a longer reference value? Is is possible to have a two week reference value? What values are you using?

Thanks!

[sodium]

Any suggestions to get a longer reference value? Is is possible to have a two week reference value? What values are you using?

the reference values depend on the values you defined in (data sources -> rra)
and there is no 1 good value: depends on your graphs (what you graph) and how steady the values are... -> just test.

[deu439356]

Right now we have ~55,000 tholds built and it takes around 150 seconds for thold to complete its run every five minutes. Does that seem right to you? It seems very slow to me, considering the poller is running in less time (90-110 seconds), collecting ~260,000 data sources.

Thanks!

[sodium]

the numbers are in the right ballpark (we are doing 30000 in 75 seconds)
depends if all datasources checked are available, the reference point in the past is yet available (new graphs have a nan for a reference point of a day ago), the graph is still active, the poller is ready, you are running boost, etc etc etc