Proxmox host reboot takes forever

gkovacs

Renowned Member
Dec 22, 2008
516
51
93
Budapest, Hungary
I had to reboot one of my Proxmox servers (1.8 with kernel 2.6.24), it rebooted really fast but PVE doesn't start up for a looooong time, like 20 minutes.

Code:
proxmox:~# pstree
init-+-2*[iscsid]
     |-master-+-pickup
     |        `-qmgr
     |-portmap
     |-rc---S20vz---S20vz---vzctl---vzquota
     |-rpc.statd
     |-rsyslogd---3*[{rsyslogd}]
     |-sh---java---42*[{java}]
     |-sshd-+-sshd---bash---pstree
     |      `-sshd
     `-udevd
Any idea what it might be doing during that time? Can I force to start PVE?
 
maybe a scheduled filesystem check (fsck)?

No, fsck runs during boot, you can't log in while it's active. As you can see above, vzquota is running, nothing else...

btw, 2.6.24 is outdated and does not work with the current packages, move to 2.6.18 or 2.6.32.
(see http://pve.proxmox.com/wiki/Proxmox_VE_Kernel)

Can I simply do an apt-get upgrade to the new kernel? Is it expected to work without causing problems to my VE's?
 
we mainly use openvz [not many kvm's] and have found 2.6.18 to be stable.

we have gone from 2.6.24 to .32 then to .18 , and the containers worked without a problem.

to switch kernels:

aptitude update
aptitude search pve-kernel

choose the kernel , and install..

aptitude install pve-kernel-2.6.18-5-pve # the latest 2.6.18 at this point.

you probably know that in /boot/grub/menu.lst the default line may need to be changed.

on reboot make sure the new kernel is selected .....
 
@ bread-baker: just use the virtual package to get the new kernel.

Code:
apt-get install proxmox-ve-2.6.18
 
2.6.18 is the only stable OpenVZ kernel and very maintained, MUCH newer than the 2.6.24. just looking on the number is not enough.
 
I switched to 2.6.18 after reading posts on this and openvz forums regarding what others thought was the best kernel for openvz. that was 9 months ago.

we were having network issues at the time. those we noticed in keyboard / data entry lags.

the cause of the problem could have been something else besides the kernel. but i just wanted to eliminate a possibility.

I made other changes like running backup scripts from the host in order to use ' ionice -c 3' .

Most of the network issues we had were solved by using managed switches.
 
2.6.18 is the only stable OpenVZ kernel and very maintained, MUCH newer than the 2.6.24. just looking on the number is not enough.

Is there some documentation that would help me understand the differences between 2.6.18 and 2.6.32?
I would like to know what features would I miss, especially if .18 is backwards in ANY way compared to .24!
 
to see if .18 is in ANY way a step backwards compared to .24 - check the openvz dev site , the kernel change logs etc.

one thing about version numbers - they do not mean much.
2.6.32 will probably always be more stable then 2.6.33 .

ubuntu 10.04 more stable then 11.04 in 18 months.

2.6.18 is used by centos and openvz maintainers afaik use centos .

for phone systems stay away from asterisk 1.6 . 1.4 is a lot better.
 
2.6.18 is based on RHEL5, one of the most stable Linux kernel. CentOS is just a copy. OpenVZ uses RHEL5 kernel for their stable branch.
 
you asked for a simple answer. 2.6.18 is the best for OpenVZ, regarding stability.

the real opposite of that simple answer would be taking a look on the source code :-)

our 2.6.32 is based on the stable Debian Squeeze kernel, also great.