Hello,

My servers didn't crash during the week-end which is a good sign.
However I'm still trying to get a dump in /var/crash. For this I need an
init ramdisk that's RDAC & MPP enabled. As I previously said I am
using LSI's drivers in order to access my SAN:

http://www.lsi.com/rdac/ds4000.html

The instalation procedure is easy. The driver package listed above comes
with a SHELL script that builds a custom initrd which then gets copied in /boot:

[xxx@localhost ~]$ ls -l /boot/mpp-2.6.18-128.2.1.el5.img
-rw------- 1 root root 4049795 Jul 20 15:28 /boot/mpp-2.6.18-128.2.1.el5.img

At this point I'm not sure if it's safe use /sbin/mkdumprd along with a custom
/etc/kdump.conf that includes all the drivers in the above ramdisk image.

Do you guys have any experience at all with this kind of stuff?

Regards,

-Andrei


On Fri, Jul 17, 2009 at 2:26 PM, Matty <matty91@gmail.com> wrote:
On Fri, Jul 17, 2009 at 11:50 AM, Andrei F<frunzales@gmail.com> wrote:
> Hi,
>
> I've configured my servers as described here:
>
> http://prefetch.net/blog/index.php/2009/07/06/using-kdump-to-get-core-files-on-fedora-and-centos-hosts/
>
> When I try to start the kdump service via service kdump start, I get the
> following warnings:
>
> [root@lweb2 boot]# service kdump start
> No kdump initial ramdisk found.                            [WARNING]
> Rebuilding /boot/initrd-2.6.18-92.1.22.el5kdump.img
> Starting kdump:
> [FAILED]
>
> First of all I like the idea of automatically building an initrd image with
> kdump support, but
> I also need MPP support. Just to give you an example, this is how both
> machines are booting up:
>
> title RDAC CentOS (2.6.18-92.1.22.el5) with MPP
> root (hd0,0)
> kernel /vmlinuz-2.6.18-92.1.22.el5 ro root=/dev/VolGroup00/LogVol00
> initrd /mpp-2.6.18-92.1.22.el5.img
>
> At this point I'm wondering how to generate an initrd image with mpp & kdump
> support.
>
>
> Also /var/log/messages gives me this:
>
> Jul 17 11:42:27 lweb2 kdump: No crashkernel parameter specified for running
> kernel
> Jul 17 11:42:27 lweb2 kdump: failed to start up
>
> I assume that once the server is being rebooted with the correct kernel
> arguments like
> crashkernel=128M@16M and the correct initrd with mpp & kdump support the
> service
> should start just fine.

You are correct. The init script checks the kernel command line
(/proc/cmdline) for the crashkernel line. If it's not present, it will
fail to initialize.

- Ryan
--
http://prefetch.net
_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos