Gnome-panel occasionally hanging trying to connect to esd in edgy

Binary package hint: gnome-panel

Gnome-panel sometimes hangs for a couple of minutes when I hit System / Quit... (which should display a window with log out, lock screen etc), it happens about one time in ten tries. After a while, it resumes normally and displays the window. As this doesn't happen every time, it has been quite hard to debug... This has happened since I upgraded to edgy around knot2 and just happened now on current edgy, 2006-10-03, and I never saw the same problem before in dapper.

Backtrace from gnome-panel:
#0 0xffffe410 in __kernel_vsyscall ()
#1 0xb7644661 in connect () from /lib/tls/i686/cmov/libpthread.so.0
#2 0xb72e1d99 in esd_close () from /usr/lib/libesd.so.0
#3 0xb72e2201 in esd_open_sound () from /usr/lib/libesd.so.0
#4 0xb7e55e78 in gnome_config_get_int_with_default_ ()
   from /usr/lib/libgnome-2.so.0
#5 0xb7e55ed7 in gnome_sound_connection_get () from /usr/lib/libgnome-2.so.0
#6 0xb7e56481 in gnome_sound_play () from /usr/lib/libgnome-2.so.0
#7 0xb7e56652 in gnome_triggers_vdo () from /usr/lib/libgnome-2.so.0
#8 0xb7e56a1b in gnome_triggers_do () from /usr/lib/libgnome-2.so.0
#9 0xb7f0e8cc in gnome_client_request_save () from /usr/lib/libgnomeui-2.so.0
#10 0x0806d444 in panel_session_request_logout ()
#11 0xb7895b29 in g_cclosure_marshal_VOID__VOID ()
   from /usr/lib/libgobject-2.0.so.0
#12 0xb788879b in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#13 0xb7898b93 in g_signal_chain_from_overridden ()
   from /usr/lib/libgobject-2.0.so.0
#14 0xb789a0b7 in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#15 0xb789a279 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#16 0xb7c5e734 in gtk_widget_activate () from /usr/lib/libgtk-x11-2.0.so.0
#17 0xb7b53778 in gtk_menu_shell_activate_item ()
   from /usr/lib/libgtk-x11-2.0.so.0
#18 0xb7b54d42 in gtk_menu_shell_append () from /usr/lib/libgtk-x11-2.0.so.0
---Type <return> to continue, or q <return> to quit---
#19 0xb7b4c935 in gtk_menu_reorder_child () from /usr/lib/libgtk-x11-2.0.so.0
#20 0xb7b46aa0 in _gtk_marshal_BOOLEAN__BOXED ()
   from /usr/lib/libgtk-x11-2.0.so.0
#21 0xb7886fb9 in g_value_set_boxed () from /usr/lib/libgobject-2.0.so.0
#22 0xb788879b in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#23 0xb78991e3 in g_signal_chain_from_overridden ()
   from /usr/lib/libgobject-2.0.so.0
#24 0xb7899e7f in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#25 0xb789a279 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#26 0xb7c5a398 in gtk_widget_get_default_style ()
   from /usr/lib/libgtk-x11-2.0.so.0
#27 0xb7b3fe93 in gtk_propagate_event () from /usr/lib/libgtk-x11-2.0.so.0
#28 0xb7b41097 in gtk_main_do_event () from /usr/lib/libgtk-x11-2.0.so.0
#29 0xb79ca7ca in _gdk_events_init () from /usr/lib/libgdk-x11-2.0.so.0
#30 0xb7740802 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
#31 0xb77437df in g_main_context_check () from /usr/lib/libglib-2.0.so.0
#32 0xb7743b89 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
#33 0xb7b41514 in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0
#34 0x08061d8c in main ()

And strace from gnome-panel (started when it hanged, and the next few lines after it resumed):

connect(18, {sa_family=AF_INET, sin_port=htons(16001), sin_addr=inet_addr("127.0.0.1")}, 16) = -1 ETIMEDOUT (Connection timed out)
close(18) = 0
open("/etc/esound/esd.conf", O_RDONLY) = 18
fstat64(18, {st_mode=S_IFREG|0644, st_size=153, ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb5e4b000
read(18, "[esd]\nauto_spawn=1\nspawn_options"..., 4096) = 153
read(18, "", 4096) = 0
close(18) = 0
munmap(0xb5e4b000, 4096) = 0

So gnome-panel hung trying to connect to esound, which never answered and is not listening to 16001 (according to lsof or netstat), and after a couple of minutes timed out. esd is however running, gnomes preferences are set to autodetect and they work, and "Enable software sound mixing (ESD)" is not checked.

Grepping through ~/.gconf/ gives:

desktop/gnome/sound/%gconf.xml: <entry name="enable_esd" mtime="1145501959" type="bool" value="false">

So if that is to be trusted it should not be using esd.

Anyway, I don't know what more info to give. If there is anything needed I can try get that, however, it will take time as this is not happening every time, only when it is really, really annoying.

BTW, googling for "htons(16001) ETIMEDOUT" gives some results so I am apparently not the only one, for example https://www.redhat.com/archives/fedora-list/2005-February/msg07286.html , but no help there...