[ovs-discuss] Help determining the cause of a kernel panic...

[Jesse Gross]

On Fri, Jan 10, 2014 at 12:34 PM, Travis Wilson
<cloudtester1225 at gmail.com> wrote:
> Good day.
>
> We are new to OpenStack and Open vSwitch and have run into an issue
> where when deploying a VM instance either our Compute Node or Network
> Node consistently kernel panic. From what we can see on the screen
> when the panics occur it appears that Open vSwitch components may be
> involved but we're so new to this process that we have any ideas on
> how to solve this issue.
>
> The environment we are working with is being setup following this guide:
> https://github.com/mseknibilel/OpenStack-Grizzly-Install-Guide/blob/OVS_MultiNode/OpenStack_Grizzly_Install_Guide.rst
>
> Outside of the 192.x.x.x addresses in this example being replaced with
> out test lab's 172.31.x.x. subnet all other configuration file changes
> are identical to what is shown in the guide above.
>
> We are loading this environment in our test lab to try and match that
> of a customer environment we working with. We'd like to use the same
> versions of OpenStack and Ubuntu that our customer is using but we
> can't seem to make things work so I wanted to ask and see if I was
> missing anything obvious. The environment we are working with is
> Ubuntu 13.04 (patched with all patches as of today) with OpenStack
> Grizzly.
>
> The Open vSwitch version number is:
> ovs-vswitchd --version
> ovs-vswitchd (Open vSwitch) 1.9.0
> Compiled Mar 1 2013 14:34:38
> OpenFlow versions 0x1:0x1
>
>
> The most important parts of your bug report are the following:
>
> * What you did that make the problem appear.
> Launched an instance of Ubuntu on OpenStack.
>
> * What you expected to happen.
> Instance would launch and have network connectivity.
>
> * What actually happened.
> During Instance launch the Compute or Network node will Kernel Panic.
>
>
> The following are also handy sometimes:
>
> * The kernel version on which Open vSwitch is running (from
> /proc/version) and the
> distribution and version number of your OS (e.g. "Centos 5.0").
>
> Linux version 3.8.0-35-generic (buildd at roseapple) (gcc version 4.7.3
> (Ubuntu/Linaro 4.7.3-
> 1ubuntu1) ) #50-Ubuntu SMP Tue Dec 3 01:24:59 UTC 2013
>
> * The contents of the vswitchd configuration database (usually
> /etc/openvswitch/conf.db).
> The output of this file looks a bit large. If these are needed please
> let me know.
>
>
> * The output of "ovs-dpctl show".
> (From our compute node)
> root at compute01:~# ovs-dpctl show
> system at br-int:
> lookups: hit:0 missed:0 lost:0
> flows: 0 port 0: br-int (internal) port 3: patch-tun (patch:
> peer=patch-int) system at br-tun: lookups: hit:0 missed:0 lost:0 flows: 0
> port 0: br-tun (internal) port 2: patch-int (patch: peer=patch-tun)
> port 4: gre-1 (gre: key=flow, remote_ip=10.20.20.52)
>
> (From our network node)
> root at network:~# ovs-dpctl show system at br-int: lookups: hit:4 missed:5
> lost:0 flows: 0 port 0: br-int (internal) port 1: tap4819cfd3-5d
> (internal) port 2: qr-b823ce99-c1 (internal) system at br-ex: lookups:
> hit:1140717 missed:580946 lost:0 flows: 403 port 0: br-ex (internal)
> port 1: qg-1527bcc2-b1 (internal) port 2: eth2
>
> * A fix or workaround, if you have one.
> None that we have been able to find.We've rebuilt all machines several
> times and continue
> to experience these issues after setup is complete in each case.
>
>
> * Any other information that you think might be relevant.
> The Compute Node in our configuration is a HP DL380 G5.
> The Network Node in our configuration is a Dell 1850 with 2 Intel and
> 2 Broadcom NICs.
>
> I have some screenshots of the Kernel Panics that we have seen but I
> wasn't sure if screenshots could be included in this email so I have
> left them out. If there is a preferred way to share these please let
> me know. I am happy to share them if they may help.

Screenshots are usually a good starting point. You can send them to the list.