On Thu, 2008-05-29 at 14:23 -0400, Alfred von Campe wrote:
> Ever since I upgraded all my systems to CentOS 5.1 I have been
> getting reports from users about "all their windows disappearing". A
> little digging revealed that they meant all gnome-terminal windows.
> Since there is only one gnome-terminal process by default for all
> your open terminal windows and tabs, a crash of that process means
> losing all your terminal windows. These crashes often (but not
> always) occur over night when the system is otherwise idle.
>
> A quick Google search found a problem on Ubuntu related to VTE (the
> terminal emulator widget used by gnome-terminal) but not much else of
> interest. Has anyone on this list experienced this? I find it hard
> to believe I'm the only one. I never saw this issue while running
> CentOS 4.X on these systems...
>
> Alfred
>

I've seen extreme slow gnome terminals in the past on Cent 5, I even
downgraded the package for the terminal to make it work better.

Never had it close randomly. I have however had applications randomly
close other then the terminal.

Gregg.


_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

Alfred von Campe wrote:
Ever since I upgraded all my systems to CentOS 5.1 I have been getting
reports from users about "all their windows disappearing". A little
digging revealed that they meant all gnome-terminal windows. Since
there is only one gnome-terminal process by default for all your open
terminal windows and tabs, a crash of that process means losing all your
terminal windows. These crashes often (but not always) occur over night
when the system is otherwise idle.


A quick Google search found a problem on Ubuntu related to VTE (the
terminal emulator widget used by gnome-terminal) but not much else of
interest. Has anyone on this list experienced this? I find it hard to
believe I'm the only one. I never saw this issue while running CentOS
4.X on these systems...




How did you upgrade?

I am running CentOS-5 on many workstations that stay on all the time and
I have never had the gnome-terminal crash.


Is it possible that you have older (possibly orphaned) binaries still
installed from the upgrade process?


If this is from a NEW install, then we need more info to help.

_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

On May 29, 2008, at 14:48, Johnny Hughes wrote:


How did you upgrade?


Fresh install via kickstart. I reformatted the root and /boot
partitions, but left one user partition untouched.


Is it possible that you have older (possibly orphaned) binaries
still installed from the upgrade process?


I don't think this is possible since I reformatted the root partition
(I only have /, /boot, and /scratch partitions in addition to a swap
partition).



If this is from a NEW install, then we need more info to help.


The process ends without leaving a trace unfortunately (at least that
I can find). I did have an strace running on one system attached to
the gnone-terminal process and it finally died after 5 days or so.
Here are the last 20 lines from the log:


open("/usr/share/X11/XErrorDB", O_RDONLY) = 27
fstat64(27, {st_mode=S_IFREG|0644, st_size=37949, ...}) = 0
read(27, "! $Xorg: XErrorDB,v 1.3 2000/08/"..., 37949) = 37949
close(27) = 0
write(2, "The program 'gnome-terminal' rec"..., 592) = 592
close(18) = 0
kill(15700, SIGTERM) = 0
writev(14, [{"GIOP1215", 12}], 1) = 12
close(14) = 0
writev(15, [{"GIOP1215", 12}], 1) = 12
close(15) = 0
writev(13, [{"GIOP1215", 12}], 1) = 12
close(13) = 0
writev(11, [{"GIOP1215", 12}], 1) = 12
close(11) = 0
close(9) = 0
close(8) = 0
unlink("/tmp/orbit-kb12698/linc-3d51-0-2491c193d89b6") = 0
close(12) = 0
exit_group(1) = ?

Is there a better way than strace to get some information on this crash?

Alfred

_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

On Thu, May 29, 2008 at 12:14 PM, Alfred von Campe <alfred@von-campe.com> wrote:
>
> Fresh install via kickstart. I reformatted the root and /boot partitions,
> but left one user partition untouched.
>
> I don't think this is possible since I reformatted the root partition (I
> only have /, /boot, and /scratch partitions in addition to a swap
> partition).
>
> The process ends without leaving a trace unfortunately (at least that I can
> find). I did have an strace running on one system attached to the
> gnone-terminal process and it finally died after 5 days or so. Here are the
> last 20 lines from the log:
>

Is it possible you have some kind of time-out set, like an idle time
cut-off? If it's only happening at night, this might explain it.

I have Seamonkey crashes and strange behavior related to nautilus
windows when I try to open Netowrk Connections, but I've never seen or
heard of anything like this.

mhr
_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

On Thu, May 29, 2008 at 3:14 PM, Alfred von Campe <alfred@von-campe.com> wrote:
> write(2, "The program 'gnome-terminal' rec"..., 592) = 592

This looks interesting, but unfortunately you cut the message too
short to know what it was saying... You might find it in
/var/log/Xorg.0.log, but I wouldn't bet on that.

> Is there a better way than strace to get some information on this crash?

strace seems fine, just use some options to enhance the output you get:
-s 1024: take 1024 bytes for every string. This wouldn't have cut that one short
-tt: if you want timestamps
-f: to follow forked processes

I almost always call strace like this:

$ strace -f -tt -s 1024 -o /tmp/strace_PROCESSNAME.$$ PIDOFTHEPROCESS

Try to do it again and you might find out what gnome-terminal was
trying to tell you.

HTH,
Filipe
_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

On Thu, May 29, 2008 at 8:58 PM, Filipe Brandenburger
<filbranden@gmail.com> wrote:
> On Thu, May 29, 2008 at 3:14 PM, Alfred von Campe <alfred@von-campe.com> wrote:
>> write(2, "The program 'gnome-terminal' rec"..., 592) = 592
>
> This looks interesting, but unfortunately you cut the message too
> short to know what it was saying... You might find it in
> /var/log/Xorg.0.log, but I wouldn't bet on that.

Try ~/.xsession-errors, that's a more likely place to find the error message.

HTH,
Filipe
_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

On May 29, 2008, at 20:58, Filipe Brandenburger wrote:

strace seems fine, just use some options to enhance the output you
get:
-s 1024: take 1024 bytes for every string. This wouldn't have cut
that one short

-tt: if you want timestamps
-f: to follow forked processes


Great suggestion. I didn't know about the -s switch. I'll change
the strace to use all those options and wait for it to fail again.


Alfred

P.S. There was nothing interesting in any of the log files.
_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos

On May 29, 2008, at 20:58, Filipe Brandenburger wrote:

strace seems fine, just use some options to enhance the output you
get:
-s 1024: take 1024 bytes for every string. This wouldn't have cut
that one short

-tt: if you want timestamps
-f: to follow forked processes


I started strace with those parameters, but the strace output did not
change:


18:55:21 read(3, "13653423420
20234?3710", 32) = 32

18:55:21 write(3, "211", 8) = 8
18:55:21 read(3, "134334371377?3110
317U 10", 32) = 32

18:55:21 open("/usr/share/X11/XErrorDB", O_RDONLY) = 25
18:55:21 fstat64(25, {st_mode=S_IFREG|0644, st_size=37949, ...}) = 0
18:55:21 read(25, "! $Xorg: XErrorDB,v 1.3 2000/08/"..., 37949) = 37949
18:55:21 close(25) = 0
18:55:21 write(2, "The program 'gnome-terminal' rec"..., 592) = 592
18:55:21 close(18) = 0
18:55:21 kill(26097, SIGTERM) = 0
18:55:21 writev(14, [{"GIOP1215", 12}], 1) = 12
18:55:21 close(14) = 0
18:55:21 writev(15, [{"GIOP1215", 12}], 1) = 12
18:55:21 close(15) = 0
18:55:21 writev(13, [{"GIOP1215", 12}], 1) = 12
18:55:21 close(13) = 0
18:55:21 writev(11, [{"GIOP1215", 12}], 1) = 12
18:55:21 close(11) = 0
18:55:21 close(9) = 0
18:55:21 close(8) = 0
18:55:21 unlink("/tmp/orbit-kb12698/linc-65ef-0-4ceb1113e5b63") = 0
18:55:21 close(12) = 0
18:55:21 exit_group(1) = ?

However, this time we started the gnome-terminal process from an
xterm, and here is what appeared on the screen after the crash:


** (gnome-terminal:26095): WARNING **: No handler for control
sequence `device-control-string' defined. [there were a bunch of
these lines, I think it happened before the crash]

The program 'gnome-terminal' received an X Window System error.
This probably reflects a bug in the program.
The error was 'BadRequest (invalid request code or no such operation)'.
(Details: serial 19864803 error_code 1 request_code 0 minor_code 0)
(Note to programmers: normally, X errors are reported asynchronously;
that is, you will receive the error a while after causing it.
To debug your program, run it with the --sync command line
option to change this behavior. You can then get a meaningful
backtrace from your debugger if you break on the gdk_x_error()
function.)


I will try to follow the instructions in the output to capture a
better error next time. I should probably also recompile gnome-
terminal with symbol information...


Alfred



_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos