Closed Bug 813494 Opened 12 years ago Closed 12 years ago

bzapi bug searches using keyword slower than equivalent search using whiteboard?

Categories

(bugzilla.mozilla.org :: API, defect)

Production
defect
Not set
normal

Tracking

()

RESOLVED WORKSFORME

People

(Reporter: emorley, Unassigned)

References

Details

In bug 791758, the keyword intermittent-failure was added to all bugs that contained [(tb-)?orange] in their whiteboard, as part of our transition away from using the whiteboard. Whilst the main win was expected to be reduced DB load for b.m.o, I kind of expected the overall search times to also decrease slightly as a result - but if anything they seem 10-20% higher?! ie compare the old bzapi call: https://api-dev.bugzilla.mozilla.org/latest/bug?whiteboard=orange&include_fields=id,summary,status,resolution&summary=android and the new: https://api-dev.bugzilla.mozilla.org/latest/bug?keywords=intermittent-failure&include_fields=id,summary,status,resolution&summary=android In addition, the former call actually includes ~150 extra bugs (bugs for other products that deliberately did not have the keyword added), so it's not like the subset that need a summary match is smaller for the old call. Could this be another bug like bug 800835 where we are doing something more inefficient by using a keyword instead of whiteboard?
hrm, i can't reproduce your results. switching ctype=csv to debug=1 in your urls, i get the following (which were typical timings from multiple runs): > Old search converts to: > > https://bugzilla.mozilla.org/buglist. > cgi?short_desc=android&status_whiteboard=orange&short_desc_type=allwordssubst > r&status_whiteboard_type=allwordssubstr&columnlist=short_desc%2Cbug_status%2C > resolution%2Cbug_id&ctype=csv Execution time: 3.347201 seconds > New search converts to: > > https://bugzilla.mozilla.org/buglist. > cgi?short_desc=android&keywords=intermittent- > failure&short_desc_type=allwordssubstr&keywords_type=allwords&columnlist=shor > t_desc%2Cbug_status%2Cresolution%2Cbug_id&ctype=csv Execution time: 2.943508 seconds
note: the times i quoted are the time just the database query took, which is significantly less prone to variance/noise when compared with total page load times (ie. template generation time is noisy).
Hmm maybe just noise then. Thank you for looking anyway :-)
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.