|
|
It seems this is a pluginwrapper issue, and it's actually present all
the way back from build 1887 (and maybe further) up to current 2068,
no matter qt or gcc version and no matter the plugin (tried flash 9,
10 b1 & b2 as well as mplayer plugin). Reason I noticed this issue at
all was that the latest builds(?) stops "working" entirely when the
cap is hit, since it won't open any new connections, making it look
like a general/system connection issue rather than an issue with
opera.
What I found happened was that when a plugin had been loaded opera
would after a while open more and more pipes (guessing it's in order
to transfer data to the wrapper), even if no new content is loaded.
The rate at which new pipes are opened seems to vary a lot (from a few
every other minute to several per couple of seconds), but eventually
it hits the ulimit open files cap.
Granted I do have roughly 100 open tabs over 3 opera windows, but not
closing pipes sure seems like a bug to me.
If no plugins are enabled, open files/pipes does not increase without
reason (and really never hits the cap).
Did you report the issue and did you get a number for it? Would like
to add this information to the report.
On 8 Juli, 14:58, 0 <r...@xxxxxxxxxxxxxxxx> wrote:
> On Mon, 07 Jul 2008 17:23:58 -0000, Jorgen Grahn <grahn+n...@xxxxxxxxxxxxxx>
> wrote:
> >> $ ulimit
> >> unlimited
>
> > Are you sure?
>
> I was looking at the documentation in the wrong place,
> and came up with the similar numbers to you:
>
> $ ulimit -a
> core file size (blocks, -c) 0
> data seg size (kbytes, -d) unlimited
> scheduling priority (-e) 0
> file size (blocks, -f) unlimited
> pending signals (-i) 16379
> max locked memory (kbytes, -l) 32
> max memory size (kbytes, -m) unlimited
> open files (-n) 1024
> pipe size (512 bytes, -p) 8
> POSIX message queues (bytes, -q) 819200
> real-time priority (-r) 0
> stack size (kbytes, -s) 8192
> cpu time (seconds, -t) unlimited
> max user processes (-u) 16379
> virtual memory (kbytes, -v) unlimited
> file locks (-x) unlimited
>
> > I cannot see from a quick glance in the documentation
> > which limit a plain 'ulimit' shows, but it sure isn't
> > the open files limit, or the core file size, or ...
>
> I agree with that.
>
> > And yes, it seems like a bug. Include the lsof
> > output in the bug report, so they can see which files
> > (or sockets, etc) are involved.
>
> Will do. Thanks.
|
|