Douglas A. Tutty wrote:

On Fri, Apr 11, 2008 at 07:10:04AM -0700, Marc Shapiro wrote:

Is there any way, instead of restricting a resource, to have a command
executed when a setpoint for a given resource is reached? Say, when FF
uses 200M of virtual memory, or over 30% of CPU, a job runs which pops
up a warning message. That way you know that a problem exists and can
handle it BEFORE it gets to the point where the system is locking up.





I wouldn't worry so much about %CPU since you can just run FF nice.

Also, remember that the system should not lock up. The worst case
should be heavy swapping (Thrashing) which means that the system will
eventually respond. However, if the offending process is running niced,
at least login and bash will have a higher priority.

Yes, the system does thrash, not a complete lockup. But, if you try to
move the cursor and few times with no results it certainly seems like a
lockup. I can usually get the system to switch to a VT where I can run
top and kill the offending process, but it can take several minutes
before it will do even that. It doesn't happen often, but with X
running on three separate terminals (my wife, daughter and I each have
our own setup running on this machine) and firefox running in each of
them it does happen every now and then. My wife currently has 18 tabs
open in firefox. My daughter only has one, but it has some kids
game/education site up, all of which use flash. I usually have three or
four tabs open, although I only have one at the moment.


I have never actually used nice, before. I guess I had better do some
research.


--
Marc Shapiro
mshapiro_42@yahoo.com




--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 04/11/08 19:54, Douglas A. Tutty wrote:
[snip]
>
> You would write a script that would pole the relavant data. Probably

That's just *disgusting*.

Or I'm a dirty old man.

- --
Ron Johnson, Jr.
Jefferson LA USA

We want... a Shrubbery!!
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFIAB6bS9HxQb37XmcRApWQAJ9ch9OXFIKiVOURep4hZL qTreFqvACgpIXv
h+/fpI3rOvvH5txsN9Y/3Z8=
=YkN7
-----END PGP SIGNATURE-----


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Douglas A. Tutty wrote:
> On Thu, Apr 10, 2008 at 01:25:53PM -0700, Steve Lamb wrote:
>> After a fresh login I'll have about 40-50Mb used, 0 swap. After
a few

>> hours of FF2.x I'd be maxed on memory and pushing 2/3rds of my 256Mb of
>> swap. Close FF and most of the memory and swap go away. FF3 doesn't
>> get that high but it still pushes my machine into swap without much
>> trouble. What really gets me is after 2-3 days logged in I'll exit
>> TBird, FF3 and Pidgin and be right back where I was when I logged in.
>> Just XFCE4 loaded. 70-80Mb used, another 30-50Mb in the swap. I log
>> out of XFCE4, log back in, 40-50Mb used, 0 swap. :/
>
> Yes, it seems that, at least on Etch, there's a memory leak in the GTK2
> widget library common to both FF and XFCE (and others). The leak does
> not happen with Konqueror.
>
> Have you tried Konqueror; you don't need all of KDE.

Has anyone mentioned FF 3? It is very much leaner and meaner that FF 2;
if you can use it, it is a nice alternative. Doesn't leak memory like a
sieve. Of course, not every Debian user can use FF 3. Installing Beta
5 is not for the faint-of-heart, though.

(I will be glad when it is finally released, and the Debian team has
packaged up an Iceweasel version of it, or whatever the name will be.)


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Ron Johnson wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 04/11/08 19:54, Douglas A. Tutty wrote:
[snip]

You would write a script that would pole the relavant data. Probably


That's just *disgusting*.

Or I'm a dirty old man.



Let me pole that for a sec...


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

On Fri, Apr 11, 2008 at 07:04:23PM -0700, Marc Shapiro wrote:
> Douglas A. Tutty wrote:
> >On Fri, Apr 11, 2008 at 07:10:04AM -0700, Marc Shapiro wrote:
> >
> Yes, the system does thrash, not a complete lockup. But, if you try to
> move the cursor and few times with no results it certainly seems like a
> lockup. I can usually get the system to switch to a VT where I can run
> top and kill the offending process, but it can take several minutes
> before it will do even that. It doesn't happen often, but with X
> running on three separate terminals (my wife, daughter and I each have
> our own setup running on this machine) and firefox running in each of
> them it does happen every now and then. My wife currently has 18 tabs
> open in firefox. My daughter only has one, but it has some kids
> game/education site up, all of which use flash. I usually have three or
> four tabs open, although I only have one at the moment.
>
> I have never actually used nice, before. I guess I had better do some
> research.

Before, I scoffed at 512 MB being considered low-ram, but you're having
one box be three workstations at the same time. I suggest that you
triple the ram in this case.

One difference between FF and Konqueror is that FF keeps the rendered
pages in its own memory map while Konqueror keeps the rendered pages in
the xorg memory map. On one of my setups it matters: when you ssh in
from a light-weight box to the power box, konqueror with many tabs
thrashes the light-weight box while FF uses the ample memory on the
power box.

I wonder if in this case, you may get better performance with konq if
there is only one xorg process. I don't know since I have never run
three x sessions at once.

Doug.


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

i find that opera does ok if i start with a clean slate each time, i.e., no
tabs from former session.

tom arnall



--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 04/12/08 09:02, Hugo Vanwoerkom wrote:
> Ron Johnson wrote:
>>
>> On 04/11/08 19:54, Douglas A. Tutty wrote:
>> [snip]
>>> You would write a script that would pole the relavant data. Probably
>>
>> That's just *disgusting*.
>>
>> Or I'm a dirty old man.
>>
>
> Let me pole that for a sec...

You burned out my mind's eye on purpose, didn't you???

- --
Ron Johnson, Jr.
Jefferson LA USA

We want... a Shrubbery!!
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFIARcNS9HxQb37XmcRAgvPAKC24GAyt+CjTLhPt3dSNm KmqLIuyQCeMWNQ
/LJgmhn2zBNyZc/tVn/TFOs=
=CtlD
-----END PGP SIGNATURE-----


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

On Sat, Apr 12, 2008 at 12:23:05PM -0700, tom arnall wrote:
> i find that opera does ok if i start with a clean slate each time, i.e., no
> tabs from former session.
>

I don't do sessions at all anyway (whatever they are), i.e. I always
have a clean slate.

Doug.


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Ron Johnson wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 04/11/08 10:21, Jean-Louis Crouzet wrote:
[snip]

Flash plugin (whatever the version) is always the root cause of my


Which version? (I run Sid.)


Iceweasel lock_up (due to swapping or other...) the number do not really
matter but site such as MySpace [Music] cause this condition almost each


Well that's your problem right there!!!! :O


time I connect there. This is really painful. I put all the possible
addons/blocker/Flash versions but no success.
I tied FF, same.
I tried another computer Pentium M, 1GB Memory and same happened...
Well I'm a bit tired now, so I just kill my Iceweasel each it locks and
restart.
I think I will have a try of the suggested "ulimit -Sv 200000" because
it just looks interesting and well spotted.


- --
Ron Johnson, Jr.
Jefferson LA USA

We want... a Shrubbery!!
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFH/733S9HxQb37XmcRApdkAJ9ryD4iJgQRu2NJSd6InBANjvA/QQCdEOvW
LpJnsCfADCX6JLqMlhETJJQ=
=Azdl
-----END PGP SIGNATURE-----


I'm running lenny 2.6.24-5 (customized). The above FF behavior is
similar even if I just use xinit and run FF from xterm.
Well I think I'd rather not under your MySpace Music comment. Thanks and
have a nice day.


JL


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

try to disable the flash plugin,then suggest use 'adblock plus' .

在 2008-04-10四的 22:07 -0400，Douglas A. Tutty写�：
> On Thu, Apr 10, 2008 at 02:45:32PM -0500, Ron Johnson wrote:
> > On 04/10/08 14:35, tom arnall wrote:
> > > all of a sudden browsers have become memory hogs on my system (>95%). i've
> >
> > I don't see how a single process could use 95% of memory on a 32-bit
> > system.
>
> Easy on my P-II, which is 32-bit (with only 64 MB memory.
>
> The real question, is how much memory is it using, not the percentage.
>
> If all of real memory is used and the box hits swap hard, its called
> Thrashing.
>
>