[SOLVED] CentOS 5.5 to 5.6 Updates - Problem installing kernel-PAE

[mbt25]

Tonight I decided to update my CentOS 5.5 box to CentOS 5.6. All of the updates went well with the exception of the update of the kernel.

My current kernel is:
Linux machine 2.6.18-194.32.1.el5PAE #1 SMP Wed Jan 5 18:43:13 EST 2011 i686 i686 i386 GNU/Linux

It stalled [and remained] here when trying to update to kernel-PAE-2.6.18-238.5.1.el5:

107/218
warning: /etc/yum.conf created as /etc/yum.conf.rpmnew
Installing : kernel-PAE 108/218

Running processes associated with the update at that time were:

root@machine [/var/log]# ps auwx|grep yum
root 2481 5.6 0.6 101076 55572 pts/0 S+ 23:03 2:16 /usr/bin/python /usr/bin/yum update

firoot@machine [/var/log]# ps auwx|grep rpm
root 5144 0.0 0.0 4576 1096 pts/0 S+ 23:13 0:00 /bin/sh /var/tmp/rpm-tmp.76457 4

root@machine [/var/log]# ps auwwwx|grep mkinit
root 5146 0.0 0.0 4576 1212 pts/0 S+ 23:13 0:00 /bin/bash /sbin/new-kernel-pkg --package kernel-PAE --mkinitrd --depmod --install 2.6.18-238.5.1.el5PAE
root 5156 0.0 0.0 4832 1484 pts/0 S+ 23:13 0:00 /bin/bash --norc /sbin/mkinitrd --allow-missing -f /boot/initrd-2.6.18-238.5.1.el5PAE.img 2.6.18-238.5.1.el5PAE
root 5219 0.0 0.0 4832 760 pts/0 S+ 23:13 0:00 /bin/bash --norc /sbin/mkinitrd --allow-missing -f /boot/initrd-2.6.18-238.5.1.el5PAE.img 2.6.18-238.5.1.el5PAE

/var/tmp/rpm-tmp.76457 contained:

if [ -f /etc/sysconfig/kernel ]; then
/bin/sed -i -e 's/^DEFAULTKERNEL=kernel-smp$/DEFAULTKERNEL=kernel-PAE/' /etc/sysconfig/kernel
fi
/sbin/new-kernel-pkg --package kernel-PAE --mkinitrd --depmod --install 2.6.18-238.5.1.el5PAE || exit $?
if [ -x /sbin/weak-modules ]
then
/sbin/weak-modules --add-kernel 2.6.18-238.5.1.el5PAE || exit $?


I killed the mkinitrd processes and the update completed, with the following error:

/var/tmp/rpm-tmp.76457: line 4: 5146 Killed /sbin/new-kernel-pkg --package kernel-PAE --mkinitrd --depmod --install 2.6.18-238.5.1.el5PAE
error: %post(kernel-PAE-2.6.18-238.5.1.el5.i686) scriptlet failed, exit status 137

rpm -qa then showed it was installed, but I did not trust that it was. I removed this kernel. It did not update /etc/grub.conf, and my old kernel and associated files were still present in /boot.

So now when I run yum check-update, it shows that kernel-PAE-2.6.18-238.5.1.el5. If I try "yum install kernel-PAE-2.6.18-238.5.1.el5", it attempts to install it but hangs just like before.

Any ideas?

Thanks

Mik

[pschaff]

Have you rebooted after installing the rest of the updates? If not, I'd reboot and try again.

[mbt25]

Thanks, I rebooted late last night. I didn't wnat to try the kernel update until I was in front of the machine, which won't be til Monday.

I also saw that udevd and a couple nash -forcesilent processes were there from midnight the day before [and shouldn't have been]. So I suspect something didn't go right over the past few days. I had decided to just reboot it. That also took care of a high load [4 vs the usual 0.50] I was seeing.

I'll attempt a kernel update tomorrow. Thanks for the suggestion.

I didn't have any problems updating a couplee of other machines prior to this one, so I was baffled and alarmed when it happened on this one since I was tired and it was early morning [and I wasn't in my best thinking mood].

Mik

[pschaff]

Understood. Please keep us posted. If this is a reproducible issue it should go in the release notes.

[mbt25]

Everything turned out fine. I did a yum-update to install kernel-PAE-2.6.18-238.5.1.el5, and it installed fine. Rebooted the server afterwards and it's working fine.

So, here is how it went:

1. Originally did a "yum update" to upgrade to CentOS 5.6.
2. It hung on kernel install
3. I removed kernel-PAE-2.6.18-238.5.1.el5 since it claimed it was installed but had not finished the process
4. I rebooted
5. I did a yum-update and it installed the single remaining package [kernel-PAE-2.6.18-238.5.1.el5].
6. Rebooted
7. All is fine


When performing step #1 above [initial upgrade of CentOS], The 'nash --forcequiet' that was showing up in the processlist from 24 hours prior to me beginning the update could have something to do with the problem I was having. I don't know. I don't know enough about what nash does or why it would have been running, but possibly it was run as part of a nightly upcp (it's a server running cPanel).

Or perhaps it just didn't want to update until I rebooted after updating everything else [i.e., leaving the kernel to update last, after reboot].

At this point I don't believe this is a reproducible problem that was due solely to the 5.5-5.6 upgrade. I think it was just a fluke thing. I've upgraded 12 other [non-cPanel] machines to CentOS 5.6 with no problem as well as three other cPanel servers, all without a hitch.

Thanks for your help.

Mik

[pschaff]

Thanks for reporting back. Marking this thread [SOLVED] for posterity.

Adding to the [url=http://wiki.centos.org/Manuals/ReleaseNotes/CentOS5.6]release notes[/url] with a pointer here just in case anyone else encounters this.