Closed
Bug 574965
Opened 14 years ago
Closed 8 years ago
crash per user chart applies throttle to crash count not adu number
Categories
(Socorro :: Webapp, task)
Tracking
(Not tracked)
RESOLVED
INVALID
People
(Reporter: chofmann, Assigned: laura)
Details
I've been applying the throttle to the adu number in my crashr chart and other places ( https://wiki.mozilla.org/CrashKill/Crashr ) , and it appears the crash per user chart currently applies the throttle pct. to the crash count. I think these give similar, but not the same results. we should figure out which one we want to use and standardize.
Reporter | ||
Comment 1•14 years ago
|
||
this would also be good to try and get for the next round of socorro development.
Assignee | ||
Comment 2•14 years ago
|
||
I'm trying to understand why we would apply it to the ADU number instead - can you enlighten me?
Reporter | ||
Comment 3•14 years ago
|
||
Here is the thinking. for a given day we have X number of users from the metrics Y number of processed crash reports in socorro Z number of raw crash reports And the possible formulas from this could be z * 0.10 / x -- assumes we are using 10% of all reports received y * 0.10 / x -- assumes we are using 10% of the number of processed reports y / x * 0.10 -- assumes we are using 10% of the number of users the first formula might be best since that reflects the server side throttling that we are actually doing, but the problem is that I don't thing we track that number or make it accessible. we would have to process the reports to get firefox version numbers out of the reports, and this won't be done until we start processing all the reports and getting rid of throttling. So this is out as an option. The way I've been thinking about this is that we want to make the assumption in throttling that we are getting a random sample of crashes from a random sample of users when we do throttling of raw incoming reports when we either do client or server side throttling. Thus, Y, or our processed data, is really just a sample a percentage of our total active users. When I talked to ken about this several months ago it made sense to him.
Updated•12 years ago
|
Component: Socorro → General
Product: Webtools → Socorro
Assignee | ||
Updated•12 years ago
|
Component: General → Webapp
Assignee | ||
Comment 4•11 years ago
|
||
Let's check we're consistent and if we are, wontfix
Assignee: nobody → laura
Comment 5•8 years ago
|
||
Crashes per User has been re-written to be based in SuperSearch.
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → INVALID
You need to log in
before you can comment on or make changes to this bug.
Description
•