All users were logged out of Bugzilla on October 13th, 2018

Add buffer tunable for writing to disk

RESOLVED INCOMPLETE

Status

()

--
enhancement
RESOLVED INCOMPLETE
10 years ago
7 years ago

People

(Reporter: yaneurabeya, Unassigned)

Tracking

(Depends on: 1 bug)

Firefox Tracking Flags

(Not tracked)

Details

(Reporter)

Description

10 years ago
User-Agent:       Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.9.0.6) Gecko/2009021414 Firefox/3.0.6
Build Identifier: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.9.0.6) Gecko/2009021414 Firefox/3.0.6

I'm comparing and contrasting performance items between Firefox and other apps, and I'm noting that the sounds of my disk (my only empirical means of measurement so far) is writing far more frequently, when other apps are idle, than Azureus is at the same transfer rates. The one thing that separates these apps are the buffer settings within Azureus are set appropriate to the amount of memory on my system.

If there was a way for downloading related items to be tuned specifically to how a user deemed it correct (via the about:config dialog for instance), that would be wonderful. I think it would make my noisy Seagate drive much happier than it is right now :).

Reproducible: Always

Comment 1

10 years ago
how many tabs do you have?

have you checked top/prstat to see how much ram/vm/swap you're using?

have you considered that this is not http://support.mozilla.com
(Reporter)

Comment 2

10 years ago
1. This isn't linux.
2. I do have a number of tabs open at once (7~20).
3. Memory isn't an issue:

[gcooper@orangebox /usr/home/gcooper]$ vmstat 
 procs      memory      page                    disks     faults         cpu
 r b w     avm    fre   flt  re  pi  po    fr  sr ad6 ad8   in   sy   cs us sy id
 0 0 0   2305M   565M   212   0   0   0   282   0   0   0  128 1056 1347  1  0 99

top(1) output [first 5 lines]:

last pid:  5723;  load averages:  0.00,  0.05,  0.15    up 1+07:34:41  05:08:36
166 processes: 1 running, 165 sleeping
CPU:  0.4% user,  0.0% nice,  0.0% system,  0.1% interrupt, 99.5% idle
Mem: 498M Active, 1982M Inact, 219M Wired, 4928K Cache, 112M Buf, 558M Free
Swap: 2048M Total, 2048M Free

Why is an enhancement considered a support request?
(Reporter)

Comment 3

10 years ago
Also, I may have accidentally been using firefox 2.x at the time (2.x and 3.x are installed on the system because of ports dependencies), and 2.x is HORRIBLE in terms of memory usage.

3.x is still a bit hungry for RAM, but not nearly as much as 2.x...
(Reporter)

Comment 4

10 years ago
Also, what I'm suggesting is to tune the value (and in my case _increase_ the buffer) because I'm not running a fairly bleeding edge PC with a fair amount of RAM and a decent processor, and having a larger buffer would be more beneficial for my system and would reduce kernel interrupts, and using mmap(3), one could better tailor firefox for more contemporary broadband connections and hardware [similar I suppose to the analogy of IPv6's larger possible MTU with (giant pages?) to IPv4's smaller fixed MTU].

Comment 5

10 years ago
well, page loads in firefox 3.0.x will generally result in a database write for places. the next major release after 3.0 will include a fix which should improve this.

but i'm not sure about idle. you could truss or something and figure out which files we're writing.

and no. a generic "something, i have no idea what, is writing to my disk, please give me a tunable" is not an enhancement, it's a support request, and a poorly worded one at that.

and tunables in general give us a larger bug surface.

if you identify a certain file or class of files - not "all disk files", and can show that it makes sense for us to provide a tunable (highly unlikely), then this is an enhancement.

until then, you're really asking for support, and the right person to help you should be someone who will have patience, and be polite to you.

i'm not sure why you said "this isn't linux", prstat is a solaris command, i could have used rosetta, and maybe support would have.

note that session restore will also hit the disk occasionally, because it wants you to be happy in case your system or firefox crashes. but again, w/o knowing which file is being touched, ....

Updated

7 years ago
Status: UNCONFIRMED → RESOLVED
Last Resolved: 7 years ago
Depends on: 729165
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.