Hi, I have updated 8 machines so far and 2 are refusing to boot on 2.6.18-194.3.1.el5 kernel, they just hang at: 'Starting udev:' but when I revert to 2.6.18-164.15.1.el5 kernel they boot fine. Any pointers?
Cheers, Colin.
Just updated ~15 box x86_64 here to CentOS 5.5. Kernel 2.6.18-194.3.1.el5 works without any problem. Thanks CentOS team for this excellent work!
On Sat, May 15, 2010 at 13:27, Colin Coles colin@wemoto.com wrote:
Hi, I have updated 8 machines so far and 2 are refusing to boot on 2.6.18-194.3.1.el5 kernel, they just hang at: 'Starting udev:' but when I revert to 2.6.18-164.15.1.el5 kernel they boot fine. Any pointers?
Cheers, Colin.
On 05/15/2010 12:27 PM, Colin Coles wrote:
I have updated 8 machines so far and 2 are refusing to boot on 2.6.18-194.3.1.el5 kernel, they just hang at: 'Starting udev:' but when I revert to 2.6.18-164.15.1.el5 kernel they boot fine. Any pointers?
Do you have anything non-standard on the machines ? Start by looking at the h/w specs and setup.
- KB
On Saturday 15 May 2010 12:55, Karanbir Singh wrote:
On 05/15/2010 12:27 PM, Colin Coles wrote:
I have updated 8 machines so far and 2 are refusing to boot on 2.6.18-194.3.1.el5 kernel, they just hang at: 'Starting udev:' but when I revert to 2.6.18-164.15.1.el5 kernel they boot fine. Any pointers?
Do you have anything non-standard on the machines ? Start by looking at the h/w specs and setup.
Thanks Karanbir, Very generic, basic off the shelf workstation stuff, both have been running CentOS for over a year without problems and other machines that are OK have very similar specs all are AMD 64:
AMD Athlon(tm) 64 X2 Dual Core Processor 6400+ 4Gb ram
AMD Athlon(tm) 64 X2 Dual Core Processor 4200+ 2 Gb ram
This is very repeatable and nothing seems to be written to /var/log/messages during failed boots, is this normal?
Cheers, Colin.
On Sat, May 15, 2010 at 4:27 AM, Colin Coles colin@wemoto.com wrote:
Hi, I have updated 8 machines so far and 2 are refusing to boot on 2.6.18-194.3.1.el5 kernel, they just hang at: 'Starting udev:' but when I revert to 2.6.18-164.15.1.el5 kernel they boot fine. Any pointers?
When I ran into a similar problem on a 5.4 system, the hints at
http://bugs.centos.org/view.php?id=2141
were helpful, particularly about increasing the udev logging level.
On Sat, 2010-05-15 at 12:27 +0100, Colin Coles wrote:
Hi, I have updated 8 machines so far and 2 are refusing to boot on 2.6.18-194.3.1.el5 kernel, they just hang at: 'Starting udev:' but when I revert to 2.6.18-164.15.1.el5 kernel they boot fine. Any pointers?
I just updated a machine and also got a hang at udev. When I revert to kernel 2.6.18-164 it works fine. It appears to be a hard hang. Caps lock and num lock have no effect and I have to hit the reset button.
I tried increasing udev log level to debug and it did not hang. I reduced the log level to info and it still did not hang. However, when I start X, esd seems to get stuck in a loop on a short segment of audio. If I kill esd, the desktop will finish initializing. Logging out of X also hung, but killing X manually worked. I have reverted to the old kernel.
Linus
On Sun, 2010-05-16 at 20:10 -0400, C Linus Hicks wrote:
On Sat, 2010-05-15 at 12:27 +0100, Colin Coles wrote:
Hi, I have updated 8 machines so far and 2 are refusing to boot on 2.6.18-194.3.1.el5 kernel, they just hang at: 'Starting udev:' but when I revert to 2.6.18-164.15.1.el5 kernel they boot fine. Any pointers?
I just updated a machine and also got a hang at udev. When I revert to kernel 2.6.18-164 it works fine. It appears to be a hard hang. Caps lock and num lock have no effect and I have to hit the reset button.
I tried increasing udev log level to debug and it did not hang. I reduced the log level to info and it still did not hang. However, when I start X, esd seems to get stuck in a loop on a short segment of audio. If I kill esd, the desktop will finish initializing. Logging out of X also hung, but killing X manually worked. I have reverted to the old kernel.
--- I am just wondering if any of you guys with the udev hang problem have tried: rpm -e the new kernel? Then try to reinstall it via yum install. You should delete the new kernel from /var/cache/yum first.
Just a point in why I say that is I have seen several machines that have yummed corrupt packages and bad repo metadata. After pegging yum to pull from a distinct mirror the problem has gone away.
John
On Mon, 2010-05-17 at 01:20 -0400, JohnS wrote:
I am just wondering if any of you guys with the udev hang problem have tried: rpm -e the new kernel? Then try to reinstall it via yum install. You should delete the new kernel from /var/cache/yum first.
Just a point in why I say that is I have seen several machines that have yummed corrupt packages and bad repo metadata. After pegging yum to pull from a distinct mirror the problem has gone away.
John
How about this: I downloaded the new kernel rpm again manually then used rpm2cpio to load the files into a work directory then ran this script:
$ rpm -qlp ../kernel-2.6.18-194.3.1.el5.x86_64.rpm | while read rl do rstr=$(cmp $rl .$rl 2>&1) rc=$? if (( rc != 0 )); then echo $rstr | grep -v "Is a directory" > /dev/null rc=$? if (( rc == 0 )); then echo "Error: $rstr" fi fi done Error: cmp: ./boot/initrd-2.6.18-194.3.1.el5.img: No such file or directory Error: cmp: ./lib/modules/2.6.18-194.3.1.el5/build: No such file or directory Error: cmp: ./lib/modules/2.6.18-194.3.1.el5/source: No such file or directory
From: C Linus Hicks linush@verizon.net
How about this: I downloaded the new kernel rpm again manually then used rpm2cpio to load the files into a work directory then ran this script: ... Error: cmp: ./boot/initrd-2.6.18-194.3.1.el5.img: No such file or directory Error: cmp: ./lib/modules/2.6.18-194.3.1.el5/build: No such file or directory Error: cmp: ./lib/modules/2.6.18-194.3.1.el5/source: No such file or directory
Maybe rpm scripts created these...?
JD
On Wed, 2010-05-19 at 18:11 -0400, C Linus Hicks wrote:
On Mon, 2010-05-17 at 01:20 -0400, JohnS wrote:
I am just wondering if any of you guys with the udev hang problem have tried: rpm -e the new kernel? Then try to reinstall it via yum install. You should delete the new kernel from /var/cache/yum first.
Just a point in why I say that is I have seen several machines that have yummed corrupt packages and bad repo metadata. After pegging yum to pull from a distinct mirror the problem has gone away.
John
How about this: I downloaded the new kernel rpm again manually then used rpm2cpio to load the files into a work directory then ran this script:
$ rpm -qlp ../kernel-2.6.18-194.3.1.el5.x86_64.rpm | while read rl do rstr=$(cmp $rl .$rl 2>&1) rc=$? if (( rc != 0 )); then echo $rstr | grep -v "Is a directory" > /dev/null rc=$? if (( rc == 0 )); then echo "Error: $rstr" fi fi done Error: cmp: ./boot/initrd-2.6.18-194.3.1.el5.img: No such file or directory Error: cmp: ./lib/modules/2.6.18-194.3.1.el5/build: No such file or directory Error: cmp: ./lib/modules/2.6.18-194.3.1.el5/source: No such file or directory
--- And that would be correct because there are not there. There generated on the kernel install by the the rpm header that contains the mkinitrd script in kernel-rt or kernel.spec.
John
-
C Linus Hicks -
Colin Coles -
Jim Davis -
John Doe -
JohnS -
Karanbir Singh -
Vnpenguin