boot error apic timer: any harm in upgrading centos kernel?

[chris_c_]

I'm seeing a boot error related to the timer.

Something like it can't find the timer.

Don't have exact error text.

Not sure if the divider=10 kernel parameter fixed it either.

The VirutalBox docs confirm this is an issue for my kernel version (pbx in a flash silver 1.7....) CentOS 5.5 64 Bit Kernel 2.6.18-194.3.1.e15

Any harm in upgrading the kernel to see if it cleans up the error??

would it be smart... or dumb.. to upgrade asterix also?

if yes to either... what's the proper commands ?

"12.4.3. Buggy Linux 2.6 kernel versions

The following bugs in Linux kernels prevent them from executing correctly in VirtualBox, causing VM boot crashes:

• The Linux kernel version 2.6.18 (and some 2.6.17 versions) introduced a race condition that can cause boot crashes in VirtualBox. Please use a kernel version 2.6.19 or later.
• With hardware virtualization and the I/O APIC enabled, kernels before 2.6.24-rc6 may panic on boot with the following message:
  Kernel panic - not syncing: IO-APIC + timer doesn't work! Boot with apic=debug and send a report. Then try booting with the 'noapic' option If you see this message, either disable hardware virtualization or the I/O APIC (see Section 3.4, “System settings”), or upgrade the guest to a newer kernel."

 

[Jake]

I am no genius at this stuff and it is probably very different with VirtualBox but I am running VMWare and use the "special or optimized kernel" that CentOS has for virtual machines. It is 2.6.18-92.1.22.el5vm and can be found from the CentOS repository. I can give you the link if you cannot find it. I think you should be fine in upgrading the kernel. If you are running PIAF 1.7 then you most likely have the most stable version of Asterisk. But you can use the update-source command and it will take care of it for you. Also, if VirtualBox does not need the vm version of kernel like VMWare then you can use update-source to update your kernel to the latest CentOS build and skip the manual kernel install instructions below.

To install a different kernel follow Wards instructions found here: http://www.pbxinaflash.net/vm/

When you install a different kernel it will not uninstall the existing kernel on the machine but will just add the new one. You can then use GRUB to pick which kernel you want to load at start up.

 

[chris_c_]

If you could give the link + basic details how to download/compile (I think I can edit menu.lst to add this other kernel as a selection) that'd be great.

 

[Jake]

The link to the CentOS Repository is:
http://dev.centos.org/~tru/kernel-vm/5/RPMS/

Then I just modified Wards instructions http://www.pbxinaflash.net/vm/ to download the most recent release of CentOS VM kernel. If you need additional help send me a private message.

 

[chris_c_]

Thanks Jake!

I would PM but think this could be useful to others.

The kernel made for vm's downloaded from dev.centos.org and built fine (except for a warning about the kernel being unsigned or missing an md5 key).

A problem doing that last step: rebuilding the zaptel config.

It failed because the folder /usr/src/zaptel does not exist.

A quick "locate zaptel" shows /root/nv/zaptel-update.sh but that also fails due to the missing directory.

I think zaptel is deprecated in favor of dahdi anyway so it shouldn't matter.


A few new serious looking red-colored errors on startup:

"FATAL: module dahdi not found.

+ about 8 other FATAL red errors, ending in xpp_usb.

Error: Missing /dev/dahdi!"

What should I do now ?

 

[Jake]

Dahdi needs to be recompiled for this kernel. Run "update-source" again. When it prompts you to update the kernel decline. There is an other way to recompile Dahdi and you can probably use the forum to find it but I find "update-source" is a good way for me.

Like you mentioned Zaptel is not used in the Silver PIAF build so you can ignore the errors.

 

[chris_c_]

** SOLVED ** Dahdi working, no errors on startup !!

Thanks to Jake for following up with me on this!

Dahdi now loads with all green OK status codes on bootup.

To get it working I just followed the steps to recompile dahdi for this kernel.

I just had to modify the filename version numbers to match the newest filenames of the kernel and kernel-devel source tar.gz files as hosted online....

rebuilt it using the steps listed and she's all good , no more scary red errors on bootup !

I'm posting my results on this thread now , so the next guy/gal knows that this really is the right solution....

thanks again to Jake

Chris