 kernel panic after ping to domain [message #37874] |
Thu, 29 October 2009 12:24  |
goeldi
Messages: 92
Registered: June 2006
|
Member |
|
|
Node is CentOS 5.4 with kernel 2.6.18-128.2.1.el5.028stab064.7ent
vzquota-3.0.12-1
vzctl-3.0.23-1
vzyum-2.4.0-11
vzpkg-2.7.0-18
vzrpm43-4.3.3-7_nonptl.6
vzctl-lib-3.0.23-1
vzrpm43-python-4.3.3-7_nonptl.6
vzrpm44-4.4.1-22.5
vzrpm44-python-4.4.1-22.5
vztmpl-centos-4-2.0-2
If I install the standard CentOS-4 Template I can ping to an ip and to a domainname from the CT without problem.
If I use a custom CentOS-5 precreated Template, pinging to ip works, but pinging to a domain name makes immediately a kernel panic.
The ethernet controller:
Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 02)
I tried these templates:
centos-5-i386-default-5.4-20091022
centos-5-i386-minimal
centos-5-x86
the hardware:
AMD Athlon II X2 240
Mainboard: ASRock A780FullHD
16GB RAM
I already tried kernel boot options noapic, nolapic and noacpi.
dmesg and /var/log/messages doesn't tell me anything special.
|
|
|
|
| Re: kernel panic after ping to domain [message #37875 is a reply to message #37874] |
Thu, 29 October 2009 12:44  |
khorenko
Messages: 533
Registered: January 2006
Location: Moscow, Russia
|
Senior Member |
|
|
Hi.
1) please, configure a remote console (preferably - serial console) - to collect the complete oops logs + messages printed right before the kernel panic. So, please, reproduce the problem, save logs, file a bug and post logs there.
http://wiki.openvz.org/Remote_console_setup
2) please, try other kernel flavors (smp(no suffix), PAE) - just to check if the problem is global or not.
3) (optional) If 1) is not possible for some reason, please, take a screenshot of the kernel panic message and post to to the bug. Not too much chances that this will help (cause not all info fit one screen), but might be still useful).
Good luck!
--
Konstantin
If your problem is solved - please, report it!
It's even more important than reporting the problem itself...
|
|
|
|
|
|
| Re: kernel panic after ping to domain [message #37877 is a reply to message #37876] |
Thu, 29 October 2009 13:47 |
goeldi
Messages: 92
Registered: June 2006
|
Member |
|
|
I tried it now with the non-pae/non-ent kernel and with this one it works ok:
ovzkernel-2.6.18-128.2.1.el5.028stab064.7
ovzkernel-debug-2.6.18-128.2.1.el5.028stab064.7
Since this system has 16GB RAM and 2 CPUs I will need the PAE or ENT version of kernel. With both of them the panic happens.
|
|
|
|
| Re: kernel panic after ping to domain [message #37878 is a reply to message #37877] |
Thu, 29 October 2009 14:02 |
goeldi
Messages: 92
Registered: June 2006
|
Member |
|
|
|
sorry, but it just happened with the debug kernel too. IPTables was deactivated on the node. The CT can resolve the IP address of the pinged domain (google.com), but that's it, after this the kernel panics.
|
|
|
|
| Re: kernel panic after ping to domain [message #37879 is a reply to message #37876] |
Thu, 29 October 2009 14:03 |
khorenko
Messages: 533
Registered: January 2006
Location: Moscow, Russia
|
Senior Member |
|
|
Hi again.
Unfortunately screenshot does not help at all in this case: you see, the calltrace we see there is not the first one (there are some lines from previous calltrace), and thus this calltrace is really useless.
Please, try to collect complete log. You've said SMP (that is "non-pae/non-ent" kernel) works fine, this let us hope the problem is not network related and thus network console might also work.
Hope that helps.
--
Konstantin
If your problem is solved - please, report it!
It's even more important than reporting the problem itself...
|
|
|
|
| Re: kernel panic after ping to domain [message #37949 is a reply to message #37879] |
Thu, 05 November 2009 15:02 |
goeldi
Messages: 92
Registered: June 2006
|
Member |
|
|
obviously the onboard ethernet adapter had a bug. I deactivated it in the BIOS and installed a pci card. Now only the el5 ovzkernel can boot these disks (no go with 2.6.18, pae or ent). Now a ping to an external domain works if IPTables on the Node is stopped. When IPTables runs, only pinging to ip adresses (not domains) is possible.
Here my configuration:
Node: CentOS 5.4 with 2.6.18-128.2.1.el5.028stab064.7ent
Attached is the sysctl.conf of the node, and the conf file of the container.
working nameserver is set inside of CT (212.40.0.10).
-
Attachment: 105.conf
(Size: 1.16KB, Downloaded 284 times)
-
Attachment: sysctl.conf
(Size: 1.23KB, Downloaded 302 times)
[Updated on: Thu, 05 November 2009 21:00] Report message to a moderator |
|
|
|
OpenVZ Forum
Members
Search
Help
Register
Login
Home
