OpenVZ Forum: Support » Troubleshooting VE start failure

Home » General » Support » Troubleshooting VE start failure

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

Troubleshooting VE start failure [message #32556]

Fri, 15 August 2008 15:44

mperkel
Messages: 253
Registered: December 2006

Senior Member

All of a sudden I'm having problems starting VEs. I'm not sure what changed but here's what I'm seeing,

Warning: distribution not specified default used /etc/vz/dists/default
Starting VE ...
VE is mounted
VE start failed
VE is unmounted

The VEs are 64bit Fedora 8.

What I'm wondering is - how do I get more information to track this down? I haven't rebooted or restarted the VZ process because I have some critical VEs running that I don't want to risk the possibility that they won't restart.

Junk Email Filter
http://www.junkemailfilter.com

Report message to a moderator

Re: Troubleshooting VE start failure [message #32558 is a reply to message #32556]

Fri, 15 August 2008 18:09

kenjy
Messages: 50
Registered: July 2008
Location: Mexico

Member

Use umount to unlock it:

umount /vz/root/VEID

then try to start it again.

http://kenjy.net |Live net !!!

Report message to a moderator

Re: Troubleshooting VE start failure [message #32562 is a reply to message #32556]

Sat, 16 August 2008 04:44

kir
Messages: 1645
Registered: August 2005
Location: Moscow, Russia

Senior Member

There are a lot of approaches...

First of all, check dmesg | tail for any erorrs. Then run vzctl start with --verbose.

Check VE config (/etc/vz/conf/$VEID.conf) -- there should be a usual set of parameters. From what I see you do not have OSTEMPLATE parameter specified -- not a problem since vzctl falls to default which happen to be RedHat/Fedora and this is what you're actually running but for the sake of completeness set it to fedora-8 or smth.

If that won't help, issue vzctl mount $VEID and cd to /vz/root/$VEID. There you can see the VE filesystem. It should look like a normal Linux/UNIX filesystem, i.e. stuff like /bin /lib /sbin /usr /proc etc.

Now you can chroot into it and run some checks:
chroot /vz/root/$VEID
# check all RPM packages, there should be a number of warnings but not too much
rpm -Va
# (if rpm won't start that something really bad happened to your VE)
# check /dev, there should be some basic stuff
ls -l /dev
# check logs, they might give you a clue what happened
less /var/log/messages
# check /sbin/init
ls -l /sbin/init
ldd /sbin/init # see that all libraries are resolved
# check /etc/inittab
less /etc/inittab

Kir Kolyshkin

http://static.openvz.org/userbars/openvz-developer.png

Report message to a moderator

Re: Troubleshooting VE start failure [message #32678 is a reply to message #32562]

Sat, 23 August 2008 07:40

pva0xd
Messages: 24
Registered: February 2008

Junior Member

And while we are on topic. Is it possible to see messages from initscripts? How? If not, how do you guys debug initsystems? I have some troubles with openrc (system to manages the services, startup and shutdown of a host) and I'm sure it prints me correct error message but I don't see it...

[Updated on: Sat, 23 August 2008 07:43]

Report message to a moderator

Re: Troubleshooting VE start failure [message #32681 is a reply to message #32556]

Sat, 23 August 2008 11:38

mperkel
Messages: 253
Registered: December 2006

Senior Member

I have to agree. I would like a vzctl option where I can start a VE and watch the init script run.

Junk Email Filter
http://www.junkemailfilter.com

Report message to a moderator

Re: Troubleshooting VE start failure [message #32684 is a reply to message #32678]

Sat, 23 August 2008 21:37

piavlo
Messages: 159
Registered: January 2007

Senior Member

IMHO ,in general ,each distribution should support logging the init boot process to some file.
With openrc you can set rc_logger="YES" and it should log the init boot output to /var/log/rc.log
(Similar to baselayout-1* with RC_BOOTLOG="yes" in /etc/rc.conf
and app-admin/showconsole installed.)
However it looks like (at least for me) setting rc_logger="YES" does not work with openrc and openvz combination. I'd be glad
if you can check it /var/log/rc.log is created then you set rc_logger="YES in /etc/rc.conf in VE with openrc installed, so that i could report this bug in openrc as a common problem for all openvz+openrc users.

Report message to a moderator

Previous Topic:	[solved] x86_64 Kernel: 2.6.18-92.1.1.el5.028s won't boot
Next Topic:	VE fails to startx

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Mon Aug 12 05:28:35 GMT 2024

Total time taken to generate the page: 0.03011 seconds