1451013 - Increase the threshold of low memory notifications on Windows

Status: RESOLVED INVALID

(Core :: General, enhancement)

Description

[Gabriele Svelto [:gsvelto] (PTO)]

Once bug 1451002 is fixed we can raise the low-memory notification threshold again, hopefully without hitting the issues we saw the last time we tried (bug 1301667). If we had a floating-threshold it would be best (bug 1308118) but that's a bonus compared to getting this stuff working in the first place.

Comment 1

[Gabriele Svelto [:gsvelto] (PTO)]

I've come across bug 1414150 while working on this and I'm a bit worried about the change described in bug 1414150 comment 2. I already tried raising the commit threshold value to 256 in bug 1301667 but I had to revert it because issues like bug 1306128 and bug 1307018. I think it's a good thing to raise that threshold but we should keep our eyes peeled for regressions unless we fix bug 1308118 really really fast. One upside of this is that we might see a decrease in OOM crashes. NI so that you're aware of the implications of that change.

Comment 2

[Nicholas Nethercote [inactive]]

Thank you for the info. The point of bug 1414150 was to remove the confusing and buggy double-specification of the pref's value. If you think 128 would be a better value for the commit space threshold, that's fine by me.

Comment 3

[Gabriele Svelto [:gsvelto] (PTO)]

(In reply to Nicholas Nethercote [:njn] from comment #2)
> Thank you for the info. The point of bug 1414150 was to remove the confusing
> and buggy double-specification of the pref's value. If you think 128 would
> be a better value for the commit space threshold, that's fine by me.

I'm fine with 256, 128 is so low that it's worthless and in fact I'd like to raise it further. We just have to ensure that it doesn't cause regressions for users. If nothing else this motivates me to fix the remaining bugs shortly.

Comment 4

[Marissa (Reese) Wood]

Which release are you targeting to fix this one? 
Is there a better component that general?
Thank you!

Comment 5

[Gabriele Svelto [:gsvelto] (PTO)]

I haven't got a deadline yet because we need to gather more telemetry data before we bump this value. I'm not sure if there would be a better component for this, this code is pretty self-contained and doesn't really belong to any specific module.

Comment 6

[Toshihito Kikuchi [:toshi]]

Bug 1586236 replaced threshold + polling model with memory resource notification. Resolving this as Invalid.