I'm in the process of updating all my CentOS systems from 4.4 to 4.5. One of the systems is spewing out the following error during the reboot:
INIT: version 2.85 booting INIT: PANIC: segmentation violation at 0x420! sleeping for 30 seconds
This happens so early in the boot I have no idea what I can do to debug this. I suspect it is a hardware problem and not something caused by the "yum update", but I thought I'd mention it here in case anyone else has any ideas.
Update: I originally sent this message out yesterday, but my email client used the wrong account so it didn't get posted to the list. I've since re-imaged this system (thank you Red Hat for Kickstart), but I am still curious what could have possibly caused this and any debugging techniques that can be used in this situation.
On a vaguely related note, I also have pondered the following in the past; if you observe a system booting, you'll see all the "[OK]" and "[FAILED]" messages. Is there a log somewhere where you can check later on which services were started and which passed or failed?
Thanks, Alfred
Alfred von Campe wrote:
On a vaguely related note, I also have pondered the following in the past; if you observe a system booting, you'll see all the "[OK]" and "[FAILED]" messages. Is there a log somewhere where you can check later on which services were started and which passed or failed?
the `dmesg` command will display the kernel message buffer, which includes messages prior to syslogd starting up. also look in /var/log/messages
as dmesg can eventually overflow with junk (iptables logging messages, for instance), I've been known to stuff a command like
dmesg > /var/log/dmesg.boot
into /etc/rc.d/rc.local just to keep a copy of the state of dmesg right after boot for later reference.
On Jun 1, 2007, at 14:41, John R Pierce wrote:
the `dmesg` command will display the kernel message buffer, which includes messages prior to syslogd starting up. also look in /var/ log/messages
as dmesg can eventually overflow with junk (iptables logging messages, for instance), I've been known to stuff a command like
dmesg > /var/log/dmesg.boot
into /etc/rc.d/rc.local just to keep a copy of the state of dmesg right after boot for later reference.
I believe that at this point init has just started, so that rc.local has not been called yet. But I'll give that a try next time.
Alfred
On Fri, Jun 01, 2007 at 11:41:09AM -0700, John R Pierce wrote:
Alfred von Campe wrote:
On a vaguely related note, I also have pondered the following in the past; if you observe a system booting, you'll see all the "[OK]" and "[FAILED]" messages. Is there a log somewhere where you can check later on which services were started and which passed or failed?
the `dmesg` command will display the kernel message buffer, which includes messages prior to syslogd starting up. also look in /var/log/messages
dmesg will only show kernel messages, as you say, so it won't help with checking [OK]/[FAILED].
Also, only those scripts that explicitly use syslog facilities (initlog(8), logger(1) or syslog(3)) will send output to syslog. All the other scripts won't, in Centos 5, as the lines that spawned initlog are all commented out in init.d/functions (don't know why).
as dmesg can eventually overflow with junk (iptables logging messages, for instance), I've been known to stuff a command like
dmesg > /var/log/dmesg.bootinto /etc/rc.d/rc.local just to keep a copy of the state of dmesg right after boot for later reference.
rc.sysinit already does that, to /var/log/dmesg.
What I usually do is change "/sbin/mingetty tty1" to "/sbin/mingetty --noclear tty1".
The contents of tty1 are available in /dev/vcs1.
On Fri, Jun 01, 2007 at 02:35:04PM -0400, Alfred von Campe wrote:
On a vaguely related note, I also have pondered the following in the past; if you observe a system booting, you'll see all the "[OK]" and "[FAILED]" messages. Is there a log somewhere where you can check later on which services were started and which passed or failed?
/var/log/boot.log eg May 20 13:48:06 versa syslog: syslogd startup succeeded May 20 13:48:06 versa syslog: klogd startup succeeded May 20 13:48:06 versa irqbalance: irqbalance startup succeeded May 20 13:48:06 versa portmap: portmap startup succeeded May 20 13:48:06 versa nfslock: rpc.statd startup succeeded
On Jun 1, 2007, at 15:11, Stephen Harris wrote:
/var/log/boot.log eg May 20 13:48:06 versa syslog: syslogd startup succeeded May 20 13:48:06 versa syslog: klogd startup succeeded May 20 13:48:06 versa irqbalance: irqbalance startup succeeded May 20 13:48:06 versa portmap: portmap startup succeeded May 20 13:48:06 versa nfslock: rpc.statd startup succeeded
Doh! Just what I was looking for -- thanks!
Alfred
On Fri, Jun 01, 2007 at 03:11:30PM -0400, Stephen Harris wrote:
On Fri, Jun 01, 2007 at 02:35:04PM -0400, Alfred von Campe wrote:
On a vaguely related note, I also have pondered the following in the past; if you observe a system booting, you'll see all the "[OK]" and "[FAILED]" messages. Is there a log somewhere where you can check later on which services were started and which passed or failed?
/var/log/boot.log eg May 20 13:48:06 versa syslog: syslogd startup succeeded May 20 13:48:06 versa syslog: klogd startup succeeded May 20 13:48:06 versa irqbalance: irqbalance startup succeeded May 20 13:48:06 versa portmap: portmap startup succeeded May 20 13:48:06 versa nfslock: rpc.statd startup succeeded
On my centos 5 systems (i386 and x86_64), that file is always empty. On centos 4, it's ok.
-
Alfred von Campe -
John R Pierce -
Luciano Rocha -
Stephen Harris