Closed
Bug 1358291
Opened 7 years ago
Closed 6 years ago
LinkedIn loads 2+ sec slower on Firefox compared to Chrome
Categories
(Core :: General, defect)
Core
General
Tracking
()
RESOLVED
WORKSFORME
Performance Impact | ? |
People
(Reporter: Harald, Unassigned)
References
(Blocks 1 open bug, )
Details
(Keywords: perf, Whiteboard: [platform-rel-LinkedIn])
Steps: - Log in - Open https://www.linkedin.com/feed/ (or https://www.linkedin.com/in/digitarald/) Pages load massively slower, over 2-3sec. This is confirmed by LinkedIn contacts. Profile on Mac/Nightly collected while navigating around the page: https://perfht.ml/2pkgHv0
Reporter | ||
Comment 1•7 years ago
|
||
The initial thought was that bug 1331680 causes most of the issue as LinkedIn read cookies 60+ times during page load. They fixed that and the cookie is read only once now but page load did not improve. naveed, can we confirm that 2sec+ slowness is just JS issues with Ember?
Flags: needinfo?(nihsanullah)
Reporter | ||
Comment 2•7 years ago
|
||
[Tracking Requested - why for this release]: Slow page load reported by users and confirmed by LinkedIn.
status-firefox53:
--- → affected
tracking-firefox53:
--- → ?
Comment 3•7 years ago
|
||
I can see the slowness, but I couldn't tell much from the profile. Although some of the JS function names were kind of weird -- i.prototype.compile(), e.prototype.compileStatement(), t.prototype.evaluate(). Seems like there must be some processing engine running within JS? evilpie, does your CacheIR dumper show anything interesting?
Blocks: 1299643
Flags: needinfo?(evilpies)
Comment 4•7 years ago
|
||
:sfink does TraceLogger give any other insight?
Flags: needinfo?(nihsanullah) → needinfo?(sphink)
Comment 5•7 years ago
|
||
I don't see anything super bad and I also don't get 2s load times. Actually because we fallback to the generic stub now, it's not necessarily easy to tell. I really need to fix bug 1348821, but I don't have a good approach for it. Some things that stood out to me: hasOwnProperty where the base is a primitive, in on an arguments object (both indexes and length) and SetElem with an index. Btw: Running CacheIR is really simple now, just start Nightly with CACHEIR_LOGS=1.
Flags: needinfo?(evilpies)
Comment 6•7 years ago
|
||
Can you test with 55 or 54 rather than 53? I can also ask for QE help here. It seems unlikely we'll be fixing this for 53, though if you disagree, please let me know.
status-firefox54:
--- → affected
status-firefox55:
--- → affected
Comment 7•7 years ago
|
||
Tracking for 55 to make sure this doesn't get lost in the shuffle.
tracking-firefox55:
--- → +
Comment 8•7 years ago
|
||
This did get lost. Andrei can your team have a look to see if this is still true? sfink, can you still see the perf issues at all?
status-firefox56:
--- → affected
Flags: needinfo?(andrei.vaida)
I did a layman test on my win10 laptop and it is indeed 2-3 seconds slower to load linkedin.com/feed on Nightly56 as compared to Chrome v59
Comment 10•7 years ago
|
||
I could also see this in latest Nightly 56 compared with Chrome 59, Windows 10 64bit. So this is definitely still here. Please let me know if there is something I can help with here.
Flags: needinfo?(andrei.vaida)
Reporter | ||
Comment 11•6 years ago
|
||
This doesn't reproduce for me since Quantum, as both Firefox and LinkedIn changed substantially.
Status: NEW → RESOLVED
Closed: 6 years ago
Flags: needinfo?(sphink)
Resolution: --- → WORKSFORME
Updated•2 years ago
|
Performance Impact: --- → ?
Whiteboard: [qf:investigate][platform-rel-LinkedIn] → [platform-rel-LinkedIn]
You need to log in
before you can comment on or make changes to this bug.
Description
•