Open Bug 1067846 Opened 10 years ago Updated 5 years ago

[Meta] Improve the performance and responsiveness of Treeherder's UI

Categories

(Tree Management :: Treeherder, defect, P3)

defect

Tracking

(Not tracked)

People

(Reporter: emorley, Unassigned)

References

(Depends on 10 open bugs)

Details

(Keywords: meta, perf, regression)

Filing this after the discussions we had at the work week. Breaking out of bug 1059286, since it's easier to keep one issue per bug (and that bug caused a regression, so will require more back and forth).
Keywords: meta
Summary: Treeherder UI interactions seem more laggy than TBPL → [Meta] Treeherder UI interactions seem more laggy than TBPL
Depends on: 1067861
Depends on: 1069389
Priority: P2 → P1
Depends on: 1062398
Depends on: 1076840
Depends on: 1084608
Depends on: 1086702
Depends on: 1090301
Depends on: 1090317
Depends on: 1095141
Keywords: regression
Depends on: 1099146
Depends on: 1100328
Depends on: 1104062
Depends on: 1110552
Depends on: 1110910
Depends on: 1117243
Making the summary a bit more general, since this bug has turned into the meta bug for overall UI perf/responsiveness issues.
Summary: [Meta] Treeherder UI interactions seem more laggy than TBPL → [Meta] Improve the performance and responsiveness of Treeherder's UI
Depends on: 1095213
Depends on: 1107667
No longer depends on: 1086702
Depends on: 1125229
Depends on: 1125245
Depends on: 1125248
Depends on: 1125251
No longer depends on: 1095213
Making this a P2 since the next actionable bug is bug 1125229, which is a P1.
Priority: P1 → P2
Depends on: 1134698
Depends on: 1136718
Depends on: 1140847
No longer blocks: treeherder-dev-transition
Depends on: 1158973
Depends on: 1158952
Depends on: 1159375
Depends on: 1160082
Blocks: 1112352
No longer blocks: 1112352
Depends on: 1112352
Depends on: 1162725
Depends on: 1164259
Depends on: 1164266
Depends on: 1170030
Depends on: 1170103
@sheriffs: This meta-bug is on the sheriff top issues list on the Treeherder meeting etherpad, however to make this more actionable (and so we know we're spending our time on the right perf issues), it would help if you could list which specific workflows are slow, so we can profile those? If you could file a new dep bug for each one, that would be awesome :-) Once we have those bugs filed, if you could pick the most urgent of them and add to the meeting etherpad - we can take a look :-) (Needinfoing a few people individually, since I can't needinfo the sheriffs alias - but everyone else watching the alias is welcome to do the above too :-))
Flags: needinfo?(wkocher)
Flags: needinfo?(ryanvm)
Flags: needinfo?(cbook)
I think bug 1076840 and bug 1164266 would help immensely with perceived and actual performance. While not something I commonly do, clicking the "load 50 more" button does lock up my browser for several seconds unless I'm on a beefier system. I'm wondering if there's some way of offloading the more intensive parts of fetching 50 additional sets of jobs that doesn't lock everything up until it finishes processing all of that new job data?
Flags: needinfo?(wkocher)
(In reply to Wes Kocher (:KWierso) from comment #6) > I think bug 1076840 and bug 1164266 would help immensely with perceived and > actual performance. This doesn't really answer the question of what workflows you're finding slow. :) Bug 1076840 has to do with initial and subsequent resultset loading performance. Are you saying that's still too slow for you? Bug 1164266 touches the same stuff, as well as update performance (which would probably be perceived as jank after the result sets are fully loaded). Are there still issues there as well? > While not something I commonly do, clicking the "load 50 more" button does > lock up my browser for several seconds unless I'm on a beefier system. I'm > wondering if there's some way of offloading the more intensive parts of > fetching 50 additional sets of jobs that doesn't lock everything up until it > finishes processing all of that new job data? I believe we've talked about doing this, most of the overhead just comes from inserting tons of elements into the DOM, which is rather unavoidable. It might be possible to do the JSON parsing in a web worker and post results back when done, which might solve some of the jank -- though I'm not sure it's worth the increase in complexity. I suspect not.
Flags: needinfo?(ryanvm)
Flags: needinfo?(cbook)
Priority: P2 → P3
Depends on: 1196209
Blocks: 1241984
No longer blocks: 1241984
Depends on: 1241984
Depends on: 1242905
Depends on: 1267102
Depends on: 1293233
No longer depends on: 1125245
Depends on: 1097101
Depends on: 1384255
Depends on: 1364888
Depends on: 1407377
Depends on: 1485226
Depends on: 1506062
Depends on: 1499551
Depends on: 1510280
Depends on: 1523698
Depends on: 1501984
Depends on: 1473777
Depends on: 1560657
Depends on: 1562017
No longer depends on: 1562017
You need to log in before you can comment on or make changes to this bug.