Upgrade issue

[stlsaint]

So i did the latest upgrade to new kernel following this guide. http://pve.proxmox.com/wiki/Downloads

Prior to this upgrade everything was working perfect. The upgrade took place with no errors and i rebooted and verified that i was completely upgraded. Then i went and tried to enter my containers just to be greeted by errors. Out of my four containers i can only vz into one of them. All my data is still there but i just cant access unless i take direct path into folder of container from host. Here is total errors im getting:

myserver:~# vzlist -a
CTID NPROC STATUS IP_ADDR HOSTNAME
101 4 running 192.168.1.107 tester.org
102 3 running 192.168.1.110 lucid_test.org
103 4 running 192.168.1.114 stuff.org
109 27 running 192.168.1.113 church3.org
myserver:~# vzctl enter 101
enter into CT 101 failed
Unable to open pty: No such file or directory
myserver:~# vzctl enter 102
entered into CT 102
root@myserver:/# exit
logout
exited from CT 102
myserver:~# vzctl enter 103
enter into CT 103 failed
Unable to open pty: No such file or directory
myserver:~#

My website is still working for now out of one of the containers that i cant enter but im assuming that these issues should be easy fixes. FYI...at the upgrade i was offered the change to remove my current vz.config and install the new one that came with upgrade. I opted to install new one but not before making a backup of the old one. Is this the cause of my issues??

Had to make edit here: I am able to reach container 109 and 102, but unable to reach 101 and 103

 

[vitor costa]

Try this:


vzctl exec VEID /sbin/MAKEDEV pty
vzctl exec VEID /sbin/MAKEDEV tty
vzctl enter VEID

Inside container:

1. Edit /etc/rc.sysinit

2. Coment:

#/sbin/start_udev

3. Add:

/sbin/MAKEDEV tty
/sbin/MAKEDEV pty

 

[dietmar]

at the upgrade i was offered the change to remove my current vz.config and install the new one that came with upgrade. I opted to install new one but not before making a backup of the old one. Is this the cause of my issues??

no. That behaviour seems to be unrelated to the host update. Maybe you updated the containers also?

 

[vitor costa]

I have this problem some time ago.... when migrating a CentOs machine to from fisical to virtual vz. I apply the solution above and all works nice.

 

[stlsaint]

no. That behaviour seems to be unrelated to the host update. Maybe you updated the containers also?

I had my containers all stopped before i updated.

@Vitor i will try your solution and report back on it. Thanks for reply

 

[stlsaint]

Try this:


vzctl exec VEID /sbin/MAKEDEV pty
vzctl exec VEID /sbin/MAKEDEV tty
vzctl enter VEID

Inside container:

1. Edit /etc/rc.sysinit

2. Coment:

#/sbin/start_udev

3. Add:

/sbin/MAKEDEV tty
/sbin/MAKEDEV pty

Bad news, after trying above it still doesnt work. Here is what i get:

myserver:~# vzlist -a
CTID NPROC STATUS IP_ADDR HOSTNAME
101 4 running 192.168.1.107 tester.org
102 3 running 192.168.1.110 lucid_test.org
103 4 running 192.168.1.114 stuff.org
109 25 running 192.168.1.113 church3.org
myserver:~# vzctl enter 101
enter into CT 101 failed
Unable to open pty: No such file or directory
myserver:~# vzctl exec 101 /sbin/MAKEDEV pty
myserver:~# vzctl exec 101 /sbin/MAKEDEV tty
myserver:~# vzctl enter 101
enter into CT 101 failed
Unable to open pty: No such file or directory
myserver:~#

 

[stlsaint]

So by the no replies i must assume that i have found either a new problem or a bug? What are my next options in either case?

 

[vitor costa]

What SO inside container ?

 

[dietmar]

So by the no replies i must assume that i have found either a new problem or a bug? What are my next options in either case?

Please try to find a way to reproduce the behavior (So that I can reproduce it here).

 

[stlsaint]

Please try to find a way to reproduce the behavior (So that I can reproduce it here).

Problem with that is i did a upgrade. So i would have to try and "un-upgrade" my server then upgrade again. Not seeing how thats possible. Only other suggestion i have been offered is to try and boot a different kernel.

 

[stlsaint]

UPDATE: So i edited menu.lst and booted original kernel (2.6.24-8-pve)
and i am able to access everything using vz tools but i am having some networking issues (ssh, updating, etc). But with newly released kernel for proxmox (2.6.32-3-pve) i am unable to reach two out of my four containers. Now according to proxmox site the kernel that sort of works for me is unsupported and only available due to legacy reasons. So i know that i will have to upgrade to new kernel sooner! But im having this same issue with new kernel.

 

[dietmar]

i am unable to reach two out of my four containers.

So you need to debug that. Is there any hint in the container init log?

 

[jmartin]

I experience similar problems with Ubuntu 9.10 templates (using Upstart). The templates worked fine before the upgrade to Proxmox 1.6.

So you need to debug that. Is there any hint in the container init log?

Upstart configuration had been fixed according to http://blog.bodhizazen.net/linux/openvz-ubuntu-9-10-templates/

That fix included mounting /proc as pretty much the very first thing. Now this mount fails. It looks like /proc has already been mounted elsewhere.

Does that ring a bell? Does proxmox mount /proc itself now?

Unfortunately, even when I avoid mounting /proc, I don't get much further (still debugging).

 

[dietmar]

Does that ring a bell? Does proxmox mount /proc itself now?

Yes, new vzctl mount /proc itself.

 

[jmartin]

Yes, new vzctl mount /proc itself.

And /sys as well...

If I avoid mounting /proc and /sys (or make the init scripts ignore the errors) my Ubuntu/Upstart containers boot again.