456910 - Stop using exclusive locking in places

Status: RESOLVED WONTFIX

(Toolkit :: Places, defect)

Description

[Shawn Wilsher :sdwilsh]

So, it turns out that using exclusive locking causes more fsyncs with the new sqlite (http://shawnwilsher.com/archives/175#comment-3134), so we should stop doing that.

The performance gains we got from it are likely not really there anymore (from drh), but I'm going to run some tests to see for sure.

Results will be posted in this bug.

Comment 1

[Shawn Wilsher :sdwilsh]

The difference for a few queries I've tried is so small (order of magnitude is something like 10^-6 %) that I don't think it really matters.

Comment 2

[Shawn Wilsher :sdwilsh]

Attachment: v1.0

Comment 3

[Shawn Wilsher :sdwilsh]

Pushed to mozilla-central:
http://hg.mozilla.org/mozilla-central/rev/ebafcda56f98

Comment 4

[Shawn Wilsher :sdwilsh]

Comment on attachment 340355 [details] [diff] [review]
v1.0

If we want to take bug 449443 on branch, we'll need this as well.

Comment 5

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Is this somehow related to what I mentioned in bug 452469 comment 51? Will this issue be gone with this check-in?

Comment 6

[Shawn Wilsher :sdwilsh]

Well, it'll help, but we'll still have zombie firefox processes around

Comment 7

[Shawn Wilsher :sdwilsh]

This will probably have to be backed out because of bug 457288

Comment 8

[Shawn Wilsher :sdwilsh]

We had to back this out for bug 457288

This is likely going to be a WONTFIX

Comment 9

[Marco Bonardo [:mak] (Away Apr 25 - May 5)]

i've seen drh has fixed this ticket that is probably what we need
http://www.sqlite.org/cvstrac/chngview?cn=5745
maybe you can patch our 3.6.3 amalgamation with that and use a 3.6.3a version, if drh is not releasing 3.6.4 soon to address that

Comment 10

[Shawn Wilsher :sdwilsh]

We cannot do this without hurting Tp badly