Apparently easily reproducible with xmonad wm.
2016-11-12 20:39:07,634 x_event_filter event=None/KeyRelease took 0.6ms 2016-11-12 20:39:07,634 parse_key_event(<gtk.gdk.Event at 0x3e8604bf7d8: GDK_KEY_RELEASE keyval=BackSpace>, False)=<GTKKeyEvent object, contents: {'modifiers'\ : [], 'group': 0, 'string': '', 'keyname': 'BackSpace', 'pressed': False, 'keyval': 65288, 'keycode': 22}> 2016-11-12 20:39:07,635 handle_key_action(ClientWindow(1), <GTKKeyEvent object, contents: {'modifiers': [], 'group': 0, 'string': '', 'keyname': 'BackSpace', \ 'pressed': False, 'keyval': 65288, 'keycode': 22}>) wid=1 2016-11-12 20:39:07,635 send_key_action(1, <GTKKeyEvent object, contents: {'modifiers': [], 'group': 0, 'string': '', 'keyname': 'BackSpace', 'pressed': False\ , 'keyval': 65288, 'keycode': 22}>) 2016-11-12 20:39:07,636 add_packet_to_queue(key-action ...) 2016-11-12 20:39:07,640 processing packet draw 2016-11-12 20:39:07,640 check_server_echo(0) last=True, server_ok=True 2016-11-12 20:39:07,641 process_draw 1 bytes for window 1 using mmap encoding with options={'flush': 0, 'rgb_format': 'RGB'} 2016-11-12 20:39:07,641 draw_region(514, 226, 760, 16, mmap, 1 bytes, 2280, {'flush': 0, 'rgb_format': 'RGB'}, [<function record_decode_time at 0x3e85ffb0398>\ , <function after_draw_refresh at 0x3e86046c938>]) 2016-11-12 20:39:07,641 record_decode_time(True, ) wid=1, mmap: 760x16, 0.4ms 2016-11-12 20:39:07,642 after_draw_refresh(True, ) 760x16 at 514x226 encoding=mmap, options={'encoding': 'mmap', 'flush': 0, 'rgb_format': 'RGB'} 2016-11-12 20:39:07,642 add_packet_to_queue(damage-sequence ...) 2016-11-12 20:39:07,643 do_expose_event(<gtk.gdk.Event at 0x3e8604bf7b0: GDK_EXPOSE area=[514, 226, 760, 16]>) area=gtk.gdk.Rectangle(514, 226, 760, 16) 2016-11-12 20:39:07,969 x_event_filter event=None/ButtonPress window=0xe0007b 2016-11-12 20:39:07,970 x_event_filter event=None/ButtonPress took 0.6ms 2016-11-12 20:39:07,971 pointer_modifiers(<gtk.gdk.Event at 0x3e8604bf7b0: GDK_BUTTON_PRESS x=508.00, y=331.00, button=2>)=((510, 350), [], []) 2016-11-12 20:39:07,971 _button_action(2, <gtk.gdk.Event at 0x3e8604bf7b0: GDK_BUTTON_PRESS x=508.00, y=331.00, button=2>, True) wid=1 / focus=1, device=Core \ Pointer, pointer=(510, 350), modifiers=[], buttons=[] 2016-11-12 20:39:07,972 add_packet_to_queue(button-action ...) 2016-11-12 20:39:07,973 x_event_filter event=('xpra-enter-event', None)/EnterNotify window=0xe0007b 2016-11-12 20:39:07,975 x_event_filter event=('xpra-enter-event', None)/EnterNotify took 1.3ms 2016-11-12 20:39:07,975 x_event_filter event=None/ButtonRelease window=0xe0007b 2016-11-12 20:39:07,976 x_event_filter event=None/ButtonRelease took 0.4ms The program 'Xpra' received an X Window System error. This probably reflects a bug in the program. The error was 'BadWindow (invalid Window parameter)'. (Details: serial 213159 error_code 3 request_code 38 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.)
Both client and server are running 0.17.6 on https://www.parabola.nu/.
If someone can reproduce this crash, it would be useful to capture the backtrace using wiki/Debugging.
Laptop: Lenovo X200 libreboot, resolution 1280x800, Parabola GNU/Linux, xmonad, grsecurity.
Installation: done via aur with the xpra-winswitch package and all dependencies listed there.
Induced by: after a while or a few hours typically when using the middle click keyboard button or several mouse-clicks in a row or mouse middle-click button.
GTK2 and GTK3 installed.
Changed to awesome WM and problems seems to have gone away. I still occasionally have a few white pixels on the windows though.
Since I cannot reproduce this, I won't be able to make progress without a gdb backtrace..
Cannot reproduce, might have been fixed in #276: r14704, r14638, r14007, r14722, r14765, r15031 (some of those commits undo each other...)
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/1356