Saving panel settings

[Brian J. Tarricone]

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

Paul Tansom wrote:
> On Thu, 2005-04-21 at 22:00 +0200, Olivier Fourdan wrote:
>
>>The panel saves its configuration autoatically, why would you like to
>>"force" a save?
> 
> Because in each of my recent crashes (forced reboots from a remote
> console) I've lost the changes I've made to it in spite of them having
> been in use for at least half a day. The last but one set of changes I
> secured by changing and then logging out and restarting XFCE, but the
> last lot I forgot to do that with. I think it lasted a day or two before
> the memory usage topped out and X was killed, but the panel changes were
> lost again.

The panel intentionally does not save its settings except on a clean
exit to avoid buggy panel plugins from causing the panel to not start at
all.  Before this change, you could (easily) get into a situation where
you add a panel plugin, the config gets saved, then the panel crashes.
The next time you try to start the panel, it crashes immediately and you
either have to edit the config file or clear it out entirely to get the
panel running again.  At least with the new behavior, the panel will
start again, though you'll lose any changes you've made.

As a solution, after you've made changes you want to save, just run
'pkill -USR1 xfce4-panel' from a terminal.  That will cause it to
restart cleanly, and save its config as a side-effect.

>>Regarding memory leaks, it can be a leak in the X server or in the
>>application, but Xfce won't cause leaks in other applications (ie if
>>Mozilla leaks, Xfce is not the culprit)
> 
> I wasn't actually blaming XFCE. I can watch the memory usage drop away
> when I close Mozilla, Firefox or Evolution (being the primary culprits).
> Sadly with the browsers closing windows doesn't help and the memory
> usage barely drops. There's no equivalent for Evolution so I can't test
> that, although it has just plain vanished on me a couple of times today
> - probably a glitch in the latest Debian unstable release! This problem
> has been going on for a few months now, but seems to be getting worse.
> I've fully tested the drives and memory otherwise I'd be thinking
> hardware.

Actually, I wouldn't think to blame hardware here - I can't see how an
OOM condition could be caused by bad hardware.  Granted, if you're just
seeing random crashes, and don't know that they're being caused by the
OOM killer, obviously you can't rule out hardware in that case.

Anyway, I routinely have Firefox and Thunderbird running for weeks at a
time without restarting them.  They *do* leak some amount of memory, but
never enough to cause an OOM condition.  Perhaps you should consider
increasing the size of your swap partition, or creating a swap file in
/tmp (or another location with sufficient free space).  Storage is
pretty cheap these days.  I have a gig of swap to act as overflow for my
640MB of physical RAM.  I've never seen the OOM killer at work.

	-brian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)

iD8DBQFCaDVg6XyW6VEeAnsRArFQAKC8Pmlsod01Klm3wpx37cd238MH5QCgpZfC
+PtK+Px5yAL3kz0gTFbXhQ4=
=O8IB
-----END PGP SIGNATURE-----