I frequently get this crashing error
Client (Xpra 0.14) MAC OSX Server (Xpra 0.14) CentOS 6.3
Applications/Xpra.app/Contents/Resources/lib/python/xpra/gtk_common/gtk2common.py:15: Warning: g_object_unref: assertion 'G_IS_OBJECT (object)' failed /Applications/Xpra.app/Contents/Resources/lib/python/gtk/gtkgl/widget.py:35: GtkWarning: gtk_widget_set_colormap: assertion 'GDK_IS_COLORMAP (colormap)' failed /Applications/Xpra.app/Contents/Resources/lib/python/xpra/client/gtk_base/gtk_client_window_base.py:152: GtkWarning: gdk_colormap_get_visual: assertion 'GDK_IS_COLORMAP (colormap)' failed Bus error: 10 Error in atexit._run_exitfuncs: Traceback (most recent call last): File "/usr/lib64/python2.6/atexit.py", line 24, in _run_exitfuncs func(*targs, **kargs) File "/usr/lib64/python2.6/logging/__init__.py", line 1516, in shutdown h.flush() File "/usr/lib64/python2.6/logging/__init__.py", line 762, in flush self.stream.flush() ValueError: I/O operation on closed file Error in sys.exitfunc: Traceback (most recent call last): File "/usr/lib64/python2.6/atexit.py", line 24, in _run_exitfuncs func(*targs, **kargs) File "/usr/lib64/python2.6/logging/__init__.py", line 1516, in shutdown h.flush() File "/usr/lib64/python2.6/logging/__init__.py", line 762, in flush self.stream.flush() ValueError: I/O operation on closed file
Do you have an intel video chipset? This looks like a duplicate of #563. If so, running without opengl should prevent the crash.
I have NVIDIA GeForce GT 650M 1024 MB
Graphics card.
Is this related?
Oh, I was hoping that the bug was limited to intel chipsets...
Can you post the output of the ./Xpra.app/Contents/Helpers/OpenGL_check
script please?
I have also upgraded to 0.15 beta build and still see the crash.. I have very similar symptoms like #563 (Application I am running is eclipse) very heavy on tooltips and menus.
I am currently running with opengl=no to see if I can repro the crash
Thanks for this great product and working on it. This is truly a great effort! I hope you find the time to fix this issue.
Thanks,
** (OpenGL-Check:24109): WARNING **: Trying to register gtype 'GSocketMsgFlags' as enum when in fact it is of type 'GFlags' 2014-09-27 15:00:43,409 OpenGL_accelerate module loaded 2014-09-27 15:00:43,409 Using accelerated ArrayDatatype 2014-09-27 15:00:43,422 2014-09-27 15:00:43,422 2014-09-27 15:00:43,422 OpenGL properties: 2014-09-27 15:00:43,422 * GLU extensions : 2014-09-27 15:00:43,423 * GLU version : 1.3 MacOSX 2014-09-27 15:00:43,423 * display_mode : SINGLE 2014-09-27 15:00:43,423 * gdkgl_version : 1.0 2014-09-27 15:00:43,423 * gdkglext_version : 1.2.0 2014-09-27 15:00:43,423 * has_alpha : True 2014-09-27 15:00:43,423 * opengl : 2.1 2014-09-27 15:00:43,423 * pygdkglext_version : 1.0.0 2014-09-27 15:00:43,423 * pyopengl : 3.1.0 2014-09-27 15:00:43,423 * renderer : NVIDIA GeForce GT 650M OpenGL Engine 2014-09-27 15:00:43,423 * rgba : True
Just an update. I have been running with opengel=no and no crashes so far. However, seeing this error now (hopefully will help track down the root cause of this issue )
2014-09-28 10:19:24,517 error processing draw packet Traceback (most recent call last): File "/Users/MacAdmin/gtk/inst/lib/python2.7/site-packages/xpra/client/ui_client_base.py", line 1791, in _draw_thread_loop File "/Users/MacAdmin/gtk/inst/lib/python2.7/site-packages/xpra/client/ui_client_base.py", line 1838, in _do_draw File "/Users/MacAdmin/gtk/inst/lib/python2.7/site-packages/xpra/client/client_window_base.py", line 262, in draw_region AssertionError: window 438 has no backing!
Thanks a lot for looking into this
Thanks for the no backing!
stacktrace, I am amazed that we've never seen it before. Should be fixed in r7827, will backport to v0.14.x
As for the other, much more serious Bus error: 10
crash... I still have no idea what is causing it.
Please do let us know if disabling opengl fixes this, as I suspect it might be related to opengl acceleration. I have also uploaded an updated (and hopefully improved) macosx beta build.
OpenGL=no seems to be very stable Not very fast though :(
any update regarding this issue? have you been able to reproduce it with tooltip heavy apps?
@mfawzymkh: none whatsoever, there is a bug, but it is probably in the gtkglext library, which isn't maintainted and is excruciatingly hard to build (see #533).
The only way forward that I can see is to move to GTK3 and hope that the bug has not been ported there (#640) or use native window code with opengl glue. Neither solution is very likely for the short term I am afraid.
Great! thanks for the explanation. Is moving to GTK3 part of v1.0 release by any chance? thanks again for working on such a great project!
Backport of r7827 for v0.14.x was in r7836.
Tooltips tend to have transparency and we've had issues surrounding that area recently. Can you try 0.14.10, and maybe also try to run the server with:
XPRA_ALPHA=0 xpra start ...
I don't have a mac with opengl to test with, so this ticket is going to keep moving slowly I am afraid.
Is the problem still present? Can you use the bug report tool to provide more info? Do you have an Intel 4000 chipset by any chance? (see #563 / #745)
Not heard back, closing.
See ticket:818#comment:10 - let's see how re-enabling opengl works out. (for 0.16)
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/695