Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: ap_verify
-
Labels:
-
Story Points:1
-
Epic Link:
-
Sprint:AP S21-6 (May)
-
Team:Alert Production
-
Urgent?:No
Description
Once the fake source pipeline is added to CI, pick up the output metrics and add them to the Chronograf dashboard.
Attachments
Issue Links
Activity
I take it back; there appears to be no way to quickly distinguish fake sources from natural ones. Deferring this to another ticket (DM-30349).
Krzysztof Findeisen
added a comment - - edited I take it back; there appears to be no way to quickly distinguish fake sources from natural ones. Deferring this to another ticket ( DM-30349 ). Hi Eric Bellm, can you take a look? I've got two different representations at the bottom of the "AP catalog metrics", one showing all three completeness metrics for a given visit-detector pair, and three showing a single metric by visit-detector like most of the other plots. Please let me know which you prefer, and whether in this case the y-axis should be clamped to [0%, 100%].
Krzysztof Findeisen
added a comment - Hi Eric Bellm , can you take a look? I've got two different representations at the bottom of the "AP catalog metrics", one showing all three completeness metrics for a given visit-detector pair, and three showing a single metric by visit-detector like most of the other plots. Please let me know which you prefer, and whether in this case the y-axis should be clamped to [0%, 100%] . Hi Krzysztof Findeisen, looks very good. For now I think I prefer the 3x single metric plots with lines for each visit-detector, and let's let the y-axis float rather than clamping to 0-100%.
Scope creep: modify non-Fakes metrics to not count fake sources. The Fakes metrics should already have filters for fake sources, so just invert them.