Re: [Xen-users] Data bleeding between domU's?

[Mark Williamson <mark.williamson@xxxxxxxxxxxx>]

> We've got a Xen setup using a 2.6.16.29 kernel with Xen 3.0.3 on the
> Dapper Drake. We recently rebooted the dom0 and when it came back
> online, the /etc/motd on one of the domU's was replaced with a snippet
> of a file that was contained on one of the other domU's -- and the only
> explanation we can conjure is that on reboot the data somehow bled
> between the LVM volumes containing the two domU. Are there any known
> instances of this? Corrupting /etc/motd is rather harmless, but I'd be
> much more concerned about the same type of accident corrupting something
> far more vital.
>
> I'm happy to provide any information that would help in tracking down
> this error. Just let me know what I can do. Thanks!

Were the domUs shut down cleanly before (or in the process of) your reboot of 
dom0?

I can think of two "obvious" (to me) ways this *could* happen, I guess...  One 
way is that somehow maybe a domain leaked the file data into system memory by 
not properly scrubbing some RAM it released at runtime.  The other way would 
be that the data somehow got moved between the LVM volumes.

Both of these scenarios sound a bit "weird".  Had you used the balloon driver  
(i.e. memory resizing via xm mem-set) in your guests at all?  Had you 
adjusted the LVM volumes?  Were the guests shut down cleanly?

It's really quite bizarre!  Have you seen anything else like this?  If you're 
worried about it, you might be able to use a tool like tripwire (or hack up a 
script of your own) in order to check for files with altered checksums over 
time.

Cheers,
Mark

-- 
Push Me Pull You - Distributed SCM tool (http://www.cl.cam.ac.uk/~maw48/pmpu/)

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users