invokes systemd?

[Joe Koenig]

Hi,

for quite some time now, I've noticed a strange behavior regarding
emelfm2. Whenever I close/quit emelfm2 I notice a significant
cpu-load. 'top' tells me this much:

%CPU %MEM TIME+ COMMAND
99.5 10.9 1:27.51 systemd-journal

What has emelfm2 to do with systemd? I'm just asking because I
weary of compiling from source in order to be able to
debug/backtrace. I'm afraid that's what I'll have to do anyway.

Kind regards
Joe


Archlinux

$ emelfm2 --version
emelFM2 v. 0.8.2
Licensed under the GPL
Copyright (C) 2003-2013, tooar <***@emelfm2.net>
Build date: Thu Jan 3 17:57:26 EST 2013
Build platform: GTK+ 2.24.14 Linux 3.7.1-2-ARCH/i686

[t***@onepost.net]

On Mon, 23 Sep 2013 09:18:59 +0200

Post by Joe Koenig
for quite some time now, I've noticed a strange behavior regarding
emelfm2. Whenever I close/quit emelfm2 I notice a significant
%CPU %MEM TIME+ COMMAND
99.5 10.9 1:27.51 systemd-journal
What has emelfm2 to do with systemd?

Nothing.

Regards
Tom

[Joe Koenig]

Post by t***@onepost.net
On Mon, 23 Sep 2013 09:18:59 +0200

Post by Joe Koenig
for quite some time now, I've noticed a strange behavior
regarding emelfm2. Whenever I close/quit emelfm2 I notice a
%CPU %MEM TIME+ COMMAND
99.5 10.9 1:27.51 systemd-journal
What has emelfm2 to do with systemd?

Nothing.

OK, I'm guessing you're saying that the invocation of systemd
is not within the realm of emelfm2 (it's source code).

I can relate to that. You're probably right. It's maybe being
invoked by the operating system. But it's evident (at least to me)
that, every time I close/quit emelfm2 my cpu-load rises
significantly by the 'systemd-journal' process.

Isn't it reasonable to assume a connection? Maybe someone can give
me a hint on where to start digging?

Regards
Joe

Archlinux

emelFM2 v. 0.8.2
Licensed under the GPL
Copyright (C) 2003-2013, tooar <***@emelfm2.net>
Build date: Thu Jan 3 17:57:26 EST 2013
Build platform: GTK+ 2.24.14 Linux 3.7.1-2-ARCH/i686

[Kyrill Detinov]

Post by Joe Koenig
Isn't it reasonable to assume a connection? Maybe someone can give
me a hint on where to start digging?

Look at ~/.xsession-error.
Remove (backup) ~/.config/emelfm2/cache.

[Joe Koenig]

Post by Kyrill Detinov
Look at ~/.xsession-error.
Remove (backup) ~/.config/emelfm2/cache.

My ~/.xsession-errors doesn't mention anything regarding
emelfm2 (~/.xsession-error doesn't exist).

Thanks anyway
Joe

[t***@onepost.net]

On Thu, 26 Sep 2013 22:38:59 +0200
Joe Koenig <***@gmx.de> wrote:

What does your actual journal newly contain, after your session shutdown ?

Regards
Tom

Post by Joe Koenig

Post by Kyrill Detinov
Look at ~/.xsession-error.
Remove (backup) ~/.config/emelfm2/cache.

My ~/.xsession-errors doesn't mention anything regarding
emelfm2 (~/.xsession-error doesn't exist).

[Geoff]

Installed systems obviously vary a great deal, but fwiw I run Arch and have
never seen this problem. I am a little unusual because (having used emelfm2
even before I switched to Arch almost 6 years ago, and having helped out a bit
with testing in the past), I compile my own emelfm2 rather than use the version
in the repos. I was running 0.8.1 until half an hour ago, until this thread
prompted me to get off my lazy rear and install :

emelFM2 v. 0.9.0
Licensed under the GPL
Copyright (C) 2003-2013, tooar <***@emelfm2.net>
Build date: Fri 27 Sep 08:55:38 BST 2013
Build platform: GTK+ 3.8.4 Linux 3.11.1-3-ARCH/i686

So far, I still cannot see any high cpu load or journal entries re emelfm2.

Geoff

[t***@onepost.net]

On Thu, 26 Sep 2013 20:54:52 +0200

Post by Joe Koenig

Post by t***@onepost.net
On Mon, 23 Sep 2013 09:18:59 +0200

Post by Joe Koenig
for quite some time now, I've noticed a strange behavior
regarding emelfm2. Whenever I close/quit emelfm2 I notice a
%CPU %MEM TIME+ COMMAND
99.5 10.9 1:27.51 systemd-journal
What has emelfm2 to do with systemd?

Nothing.

OK, I'm guessing you're saying that the invocation of systemd
is not within the realm of emelfm2 (it's source code).
I can relate to that. You're probably right. It's maybe being
invoked by the operating system. But it's evident (at least to me)
that, every time I close/quit emelfm2 my cpu-load rises
significantly by the 'systemd-journal' process.
Isn't it reasonable to assume a connection? Maybe someone can give
me a hint on where to start digging?

This is a long-shot guess.

In 0.8.2 at least, one-too-many data lines could sometimes get into the config data for any plugin that has more than one action e.g. the copy plugin. Because of this, at session-end the application tried to clean up the related resources twice, and crashed. Users would generally not notice the crash, as it happened late in the shutdown process.

Would that trigger a systemd hissy fit?

If so, best choice is to upgrade to 0.9.0 and refresh your plugins config data.

You could try editing your config file, look for things like -

true|true|_Copy|plugin_copy_48.png||e2p_cpbar.so||cpbar0.8.2
true|true|_Copy||Copy selected .....
true|true|Copy with _times||Copy selected .....

There should be only one indented line for each action. I don't know if manually removing any extra line will permanently fix the crash.

Regards
Tom

Post by Joe Koenig
emelFM2 v. 0.8.2
Licensed under the GPL
Build date: Thu Jan 3 17:57:26 EST 2013
Build platform: GTK+ 2.24.14 Linux 3.7.1-2-ARCH/i686

[Joe Koenig]

[ ... Whenever I close/quit emelfm2 I notice a significant
cpu-load ... ]

Post by t***@onepost.net
This is a long-shot guess.

[...]

Thanks, no matter what.

I haven't followed up on it because I wanted to debug from source.

I compiled from source with
$ export CFLAGS="-O1 -ggdb" CXXFLAGS="${CFLAGS}" DEBUG=1
PREFIX=/home/joe/tmp/emelfm2/

$ make install

After that, I launched
/home/joe/tmp/emelfm2/bin/emelfm2
a couple of times.

Then, I launched
/usr/bin/emelfm2
a couple of times, too.

Funny thing is: I couldn't reproduce the error/bug in either,
anymore. emelfm2 would just gracefully quit, as supposed. No more
cpu-peaks.

Don't know what to make of this. Wild guess: the 0.9 version I
compiled from source, fiddled with my 0.8.2 version's config file.

Thanks for your attention.
Kind regards
Joe