Opened 16 years ago
Closed 13 years ago
#3709 closed defect (fixed)
failed to create a new session error
Reported by: | zebul6 | Owned by: | |
---|---|---|---|
Component: | other | Version: | VirtualBox 2.2.0 |
Keywords: | Cc: | ||
Guest type: | other | Host type: | other |
Description
as i have no help on the forum for my problem http://forums.virtualbox.org/viewtopic.php?f=7&t=16316 my last solution is to ditch virtualbox until it works for me. i fill this bug report for the dev to know it does not work for everybody. may be it is the same bug than Ticket #3568 ?
i upgraded to virtualbox 2.2.0 PUEL and when i run now virtualbox, after it tells me to migrate my configuration files, it begins to throw me errors about 'failed to create a new session'. if i click on details i see a NS_ERROR_FACTORY_NOT_REGISTERED (0x80040154) error code.
it displays that error for each of my VM. if i click ok for each of my VM, i can get the virtualbox GUI running. but i get the same error again if i try to run one of my VM or if i try to create a new virtual drive. this errors even happen with a new user with a clean ~/.VirtualBox and no VM when i try to create a new virtual drive.
i am running 2.6.29 kernel and a self made virtualbox package which allowed to run vbox 2.1.4 sucessfully and without errors.
it seems to happen since i upgraded to kernel 2.6.29 (but other packages have been upgraded too)
running archlinux current(rolling release) as of today 2009-04-12
i checked the perm on /tmp wich are drwxrwxrwt 12 root root 72K avril 12 08:32 /tmp/
Attachments (3)
Change History (13)
comment:1 by , 16 years ago
comment:2 by , 16 years ago
Please could you provide me the following log:
sudo chmod u+s /usr/bin/strace killall VBoxSVC strace -f -s128 -o ~/log /opt/VirtualBox/VirtualBox -startvm VM_NAME
Please adapt the path to your VBox installation and use a proper VM name for VM_NAME. Please compress the output of strace and attach the resulting file to this ticket.
comment:3 by , 16 years ago
About the attachement log.bz2, i have run
strace -f -s128 -o ~/log /opt/virtualbox/VirtualBox -startvm Archlinux
a archlinux (guest) in a archlinux (host) (because it's the only VM that was completly shutdown and NOT in a saved state)
then, i got the pop-up to convert my configuration file again. i clicked OK. then i got 4 times (because i have 3 VMs) the error dialog (failed to create new session) and i clicked ok each time. it ended in console (tilda) with the message
resume: ptrace(PTRACE_SYSCALL, ...): No such process
i tried another time and i got in console this time
upeek: ptrace(PTRACE_PEEKUSER,6264,44,0): No such process resume: ptrace(PTRACE_SYSCALL, ...): No such process
note i run virtualbox 2.2.0 in gnome 2.26 with xorg-server 1.6.0 with nvidia driver 180.44 (and with qt 4.5 if it's matter)
comment:4 by , 16 years ago
Are you sure that you are running the PUEL version? Did you run any OSE version before?
Please have a look at #3672. The symlinks as described in one of my last comments are missing at your host as well (I saw that in the strace log). Add the symlinks and your VM should start fine. OTOH, these symlinks should be created by the .run installer, therefore I was asking if you really using the PUEL version.
comment:5 by , 16 years ago
yes i use the PUEL
my bad. i noticed the symlink but used by error ../.. instead of ../ it's working now because the symlinks are correct now /o\
thank you for the help ;-)
to explain in details. i wanted to have a package for virtualbox and do not want to use install.sh as root. so i patched install.sh so that it install itself in a temporary directory that i use to package all the files. (i make another package for the modules) that's where i made my error
so i take this oportunity to request a feature. allow us to install in a given directory with install.sh (with the use of DESTDIR like we can with make install) to easily package virtualbox.
i will attach the PKGBUILD i use. you might want to look at all the sed line i use to "patch" install.sh so that i could install in a given directory. i will attach too a patch which will be more easily read that the sed line
by , 16 years ago
Attachment: | DESTDIR.patch added |
---|
kind of the patch i use on install.sh to allow use of a DESTDIR variable
comment:6 by , 16 years ago
Hmm, but you can override INSTALLATION_DIR by calling install.sh with an additional parameter, see the original install.sh line 186. Isn't that sufficient?
comment:7 by , 16 years ago
been there, tried that. i guess if i have done that, they might be a reason. first /var/log/ is only writable by root not by a user. so /var/log/anything is a problem in install.sh and in uninstall.sh
then install.sh try to create files in /etc not in INSTALLATION_DIR so it does not work and there is really a need to use a DESTDIR variable.
install.sh is not designed to do that: install in a given directory to be able to package it. it simply only can install in some directory but on the live system
comment:8 by , 16 years ago
But you need to be root to install VBox properly.
Of course, install.sh is not designed to install the files for packaging. The proposed way for packaging is to use the entire bin/ directory for your package. Of course you might want to remove the testcases and stuff, please have a look at our build instructions. See r18883, r18884 for the fixes which add the proper symlinks to the bin/components directory.
comment:9 by , 16 years ago
I forgot to say I (and archlinux) use fakeroot to fake a root environment while packaging as a user.
comment:10 by , 13 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
Please reopen if still relevant.
and i tried to downgrade to 2.1.4 and i got now the exact same error. OMG
i used the backup virtualbox-1.6.xml.back config file with vbox 2.1.4
so i can't either run vbox 2.1.4. damn !