![]() |
|
Home | Help | Status | Forums | Glossary | Account
|
log in |
Message boards : Pirate Applications : cube 6.06
Author | Message |
---|---|
Version 6.05 of the cube application had a problem on at least one 64-bit machine (see other thread). I fiddled with the code and fixed a minor problem. I doubt that this fixed the real problem, but it's worth checking before we try something else, so I've released version 6.06. With St. Valentine's day long past, the graphics have returned to the simple rotating cube. | |
ID: 8834 | Rating: 0 | rate:
![]() ![]() ![]() | |
Same behavior for me. [kathryn@Nova ~]$ boincmgr boincmgr: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0. boincmgr: xcb_io.c:182: process_responses: Assertion `((int) (((dpy->last_request_read)) - ((dpy->request))) <= 0)' failed. SIGABRT: abort called Stack trace (19 frames): boincmgr(boinc_catch_signal+0x43)[0x532b33] /lib64/libpthread.so.0[0x365440f0f0] /lib64/libc.so.6(gsignal+0x35)[0x3653832f05] /lib64/libc.so.6(abort+0x183)[0x3653834a73] /lib64/libc.so.6(__assert_fail+0xe9)[0x365382bef9] /usr/lib64/libX11.so.6[0x3774c4d05e] /usr/lib64/libX11.so.6(_XEventsQueued+0x36)[0x3774c4d726] /usr/lib64/libX11.so.6(XPending+0x5d)[0x3774c35ccd] /usr/lib64/libgdk-x11-2.0.so.0[0x377884b6f5] /lib64/libglib-2.0.so.0(g_main_context_check+0x29a)[0x363503a51a] /lib64/libglib-2.0.so.0[0x363503ae71] /lib64/libglib-2.0.so.0(g_main_loop_run+0x1cd)[0x363503b4bd] /usr/lib64/libgtk-x11-2.0.so.0(gtk_main+0xa7)[0x37791238a7] boincmgr[0x654eb0] boincmgr[0x5e0544] boincmgr[0x6d5b60] boincmgr[0x44cf72] /lib64/libc.so.6(__libc_start_main+0xe6)[0x365381e576] boincmgr[0x420299] Exiting... And the graphics app when run under my account crashes as well. [kathryn@Nova tmp]$ cat stderrgfx.txt SIGSEGV: segmentation violation Stack trace (17 frames): [0x80c6f67] [0x486400] [0x8244aa] [0x821989] [0x81cc09] [0xe854ec] [0xe81b62] [0xe826ae] [0x80c09e5] [0x80c0b9e] [0x80c0de0] [0x80c0f01] [0x80c1385] [0x80bc0c4] [0x80541cd] [0x805ac1a] [0x8054101] Exiting... Tomorrow night I'll update to the latest alpha build of the CC/Manager and see if that makes a difference. And FWIW, the xphoon I downloaded did work. It's hidden under my desktop wallpaper. I could see it flash by when I locked my desktop. ____________ Kathryn :o) The BOINC FAQ Service The Unofficial BOINC Wiki The Trac System | |
ID: 8835 | Rating: 0 | rate:
![]() ![]() ![]() | |
Interesting. The manager crashing has something to do with permissions to use the display. If I su to boinc's account and then show graphics (with E@H as I don't have any P@H right now) the graphics display properly. | |
ID: 8836 | Rating: 0 | rate:
![]() ![]() ![]() | |
KSMarksPsych wrote:
Ah, that is interesting. xphoon writes to the "root" window of the display, which is behind all others. This says that the wallpaper has its own borderless window in front of the root window, instead of being written to the root window. I didn't know that; now I do. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
ID: 8837 | Rating: 0 | rate:
![]() ![]() ![]() | |
KSMarksPsych wrote: Interesting. The manager crashing has something to do with permissions to use the display. If I su to boinc's account and then show graphics (with E@H as I don't have any P@H right now) the graphics display properly. Good work. I suspected it might be something like this, but we needed proof. Now that we know this, we have to come up with a way to allow the graphics to show for users who are in the boinc group. Or grant the boinc user access using something like the MIT magic cookie. Fun. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
ID: 8838 | Rating: 0 | rate:
![]() ![]() ![]() | |
I had this behavior today, (Time : UTC+1) | |
ID: 8839 | Rating: 0 | rate:
![]() ![]() ![]() | |
KSMarksPsych wrote: Interesting. The manager crashing has something to do with permissions to use the display. If I su to boinc's account and then show graphics (with E@H as I don't have any P@H right now) the graphics display properly. So two things to try with 'xhost' are: xhost +local:boinc xhost +local: The first grants the boinc user permission to open a window in your X11 session. The second grants any local user that permission. You don't want to use `xhost +` by itself because it also allows non-local (ie. network) connections and then you can get sniffed or other troubles. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
ID: 8840 | Rating: 0 | rate:
![]() ![]() ![]() | |
littleBouncer wrote: I had this behavior today, (Time : UTC+1) For some reason it seems your machine tried to report the same WU twice. But it made it the first time, so I don't think there is a problem (unless it continues to happen). ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
ID: 8841 | Rating: 0 | rate:
![]() ![]() ![]() | |
Wormholio wrote:
No luck. I tested with E@H to remove the factor of the cube graphics app crashing. In both cases, the manager crashes, albeit with different stack dumps. [kathryn@Nova ~]$ xhost +local:boinc non-network local connections being added to access control list [kathryn@Nova ~]$ boincmgr boincmgr: Fatal IO error 13 (Permission denied) on X server :0.0. boincmgr: xcb_io.c:182: process_responses: Assertion `((int) (((dpy->last_request_read)) - ((dpy->request))) <= 0)' failed. SIGABRT: abort called Stack trace (19 frames): boincmgr(boinc_catch_signal+0x43)[0x532b33] /lib64/libpthread.so.0[0x365440f0f0] /lib64/libc.so.6(gsignal+0x35)[0x3653832f05] /lib64/libc.so.6(abort+0x183)[0x3653834a73] /lib64/libc.so.6(__assert_fail+0xe9)[0x365382bef9] /usr/lib64/libX11.so.6[0x3774c4d05e] /usr/lib64/libX11.so.6(_XEventsQueued+0x36)[0x3774c4d726] /usr/lib64/libX11.so.6(XPending+0x5d)[0x3774c35ccd] /usr/lib64/libgdk-x11-2.0.so.0[0x377884b6f5] /lib64/libglib-2.0.so.0(g_main_context_check+0x29a)[0x363503a51a] /lib64/libglib-2.0.so.0[0x363503ae71] /lib64/libglib-2.0.so.0(g_main_loop_run+0x1cd)[0x363503b4bd] /usr/lib64/libgtk-x11-2.0.so.0(gtk_main+0xa7)[0x37791238a7] boincmgr[0x654eb0] boincmgr[0x5e0544] boincmgr[0x6d5b60] boincmgr[0x44cf72] /lib64/libc.so.6(__libc_start_main+0xe6)[0x365381e576] boincmgr[0x420299] Exiting... [kathryn@Nova ~]$ xhost +local: non-network local connections being added to access control list [kathryn@Nova ~]$ boincmgr boincmgr: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0. [kathryn@Nova ~]$ boincmgr: xcb_io.c:182: process_responses: Assertion `((int) (((dpy->last_request_read)) - ((dpy->request))) <= 0)' failed. SIGABRT: abort called Stack trace (19 frames): boincmgr(boinc_catch_signal+0x43)[0x532b33] /lib64/libpthread.so.0[0x365440f0f0] /lib64/libc.so.6(gsignal+0x35)[0x3653832f05] /lib64/libc.so.6(abort+0x183)[0x3653834a73] /lib64/libc.so.6(__assert_fail+0xe9)[0x365382bef9] /usr/lib64/libX11.so.6[0x3774c4d05e] /usr/lib64/libX11.so.6(_XEventsQueued+0x36)[0x3774c4d726] /usr/lib64/libX11.so.6(XPending+0x5d)[0x3774c35ccd] /usr/lib64/libgdk-x11-2.0.so.0[0x377884b6f5] /lib64/libglib-2.0.so.0(g_main_context_check+0x29a)[0x363503a51a] /lib64/libglib-2.0.so.0[0x363503ae71] /lib64/libglib-2.0.so.0(g_main_loop_run+0x1cd)[0x363503b4bd] /usr/lib64/libgtk-x11-2.0.so.0(gtk_main+0xa7)[0x37791238a7] boincmgr[0x654eb0] boincmgr[0x5e0544] boincmgr[0x6d5b60] boincmgr[0x44cf72] /lib64/libc.so.6(__libc_start_main+0xe6)[0x365381e576] boincmgr[0x420299] Exiting... ____________ Kathryn :o) The BOINC FAQ Service The Unofficial BOINC Wiki The Trac System | |
ID: 8842 | Rating: 0 | rate:
![]() ![]() ![]() | |
I'll also add I just upgraded to 6.10.36 (the latest Linux alpha) and the manager still crashes. | |
ID: 8843 | Rating: 0 | rate:
![]() ![]() ![]() | |
This is on Fedora 10, right? Because I also tested it on Fedora 10 and it worked for me. | |
ID: 8844 | Rating: 0 | rate:
![]() ![]() ![]() | |
Just to add what we are learning, if I run the cube graphics program manually from the 'boinc' user account with xhost turned off so that it should not be able to open a window, then the following appears in stderrgfx.txt: No protocol specified GLUT: Fatal Error in cube_graphics_6.05_i686-pc-linux-gnu: could not open displa y: :0 SIGSEGV: segmentation violation That's less cryptic than what Kathryn has reported. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
ID: 8845 | Rating: 0 | rate:
![]() ![]() ![]() | |
Wormholio wrote: This is on Fedora 10, right? Because I also tested it on Fedora 10 and it worked for me. Yes. 64 bit Fedora 10. KDE as the desktop, FWIW. You're not using any of those fancy smancy window managers, are you? :) Per Rom's message to the alpha list, I checked if it does the same without fiddling with the xhost permissions and it does. The manager crashes just as hard. I rebooted because I didn't now how to change the xhost stuff back to default. [kathryn@Nova ~]$ boincmgr boincmgr: xcb_io.c:182: process_responses: Assertion `((int) (((dpy->last_request_read)) - ((dpy->request))) <= 0)' failed. SIGABRT: abort called Stack trace (19 frames): boincmgr(boinc_catch_signal+0x43)[0x534283] /lib64/libpthread.so.0[0x365440f0f0] /lib64/libc.so.6(gsignal+0x35)[0x3653832f05] /lib64/libc.so.6(abort+0x183)[0x3653834a73] /lib64/libc.so.6(__assert_fail+0xe9)[0x365382bef9] /usr/lib64/libX11.so.6[0x3774c4d05e] /usr/lib64/libX11.so.6(_XEventsQueued+0x36)[0x3774c4d726] /usr/lib64/libX11.so.6(XPending+0x5d)[0x3774c35ccd] /usr/lib64/libgdk-x11-2.0.so.0[0x377884b6f5] /lib64/libglib-2.0.so.0(g_main_context_check+0x29a)[0x363503a51a] /lib64/libglib-2.0.so.0[0x363503ae71] /lib64/libglib-2.0.so.0(g_main_loop_run+0x1cd)[0x363503b4bd] /usr/lib64/libgtk-x11-2.0.so.0(gtk_main+0xa7)[0x37791238a7] boincmgr[0x656670] boincmgr[0x5e1d04] boincmgr[0x6d7320] boincmgr[0x44d0a2] /lib64/libc.so.6(__libc_start_main+0xe6)[0x365381e576] boincmgr[0x420299] Exiting... [kathryn@Nova ~]$ boincmgr: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0. ____________ Kathryn :o) The BOINC FAQ Service The Unofficial BOINC Wiki The Trac System | |
ID: 8846 | Rating: 0 | rate:
![]() ![]() ![]() | |
KSMarksPsych wrote: Yes. 64 bit Fedora 10. KDE as the desktop, FWIW. You're not using any of those fancy smancy window managers, are you? :) It worked correctly for me on Fedora 10 without KDE or Gnome. I thought I had the problem with Ubuntu and Gnome, but then it worked with Einstein and I'm waiting for more work from Pirates (someone should do something about that :-) I also have a machine with KDE, but I upgraded this morning from F8 to F10 and now it's hosed, so no test there until I sort that out. You can say `xhost -` to turn off all access, and then add to that. Or Control-Alt-Backspace (not Delete) will restart the X server. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
ID: 8847 | Rating: 0 | rate:
![]() ![]() ![]() | |
I think I've been able to replicate this problem. | |
ID: 8848 | Rating: 0 | rate:
![]() ![]() ![]() | |
Wormholio wrote:
A little late to the party, but I had this happening continually. It gave me time to verify what was causing the error. Basically, BOINC would do standard communication with the Pirate's server and time out after a period of no response. Since the results were not removed, this error appeared the next time the manager tried to communicate. On occasion, the first communication would finish normally, but the manager would not remove the tasks also causing the error. Between the two, I would get an error as much as a third of the time some days. ____________ | |
ID: 8849 | Rating: 0 | rate:
![]() ![]() ![]() | |
KWSN - A Shrubbery, arrr wrote: Basically, BOINC would do standard communication with the Pirate's server and time out after a period of no response. Since the results were not removed, this error appeared the next time the manager tried to communicate. On occasion, the first communication would finish normally, but the manager would not remove the tasks also causing the error. I have noticed during this run that the database server can get really bogged down. I've tried restarting it, but that only helped a little. And I notice from the status page (which can be really slow to load) that there are over 4,000 "messages from hosts" -- trickle messages. The only thing is, this application does not send or receive trickle messages. This may or may not be the cause of the slow-down, but it certainly needs to be explained. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
ID: 8850 | Rating: 0 | rate:
![]() ![]() ![]() | |
I flushed out all the old trickle messages and restarted the database server, web server, and project. That seems to have improved performance, but the large number of workunits in the database may also still be an issue. We don't have a very powerful database server. We will see how it goes as what is now in the database is purged. | |
ID: 8851 | Rating: 0 | rate:
![]() ![]() ![]() | |
Next one going to be green shamrocks? :-) | |
ID: 8852 | Rating: 0 | rate:
![]() ![]() ![]() | |
Message boards : Pirate Applications : cube 6.06
Home | Help | Status | Forums | Glossary | Account
|