[xenpreview-dev] SUSE Xen 9762 live migration tests

[John Byrne]

Martin Scholl wrote:
> "Hello" to all as this is my first posting here...

Well, it looks like you didn't to cc the list so I'm not sure it counts.

> 
> 
> John Byrne schrieb:
>> I ran a "ping-pong" live migration test on x86-64 of a 3GB UP guest
>> domain running kernel makes over the weekend. (I was using a shared disk
>> for the storage.) 2664 migrations (the script had a 2 minute pause

I just found that this should have been 1332 loops.

>> between migrations): no hangs, no panics, no obvious errors to be found
>> in the output from the make. fsck found no corruptions on the disk.
> Wow!
> 
> Were any stress-tests performed while testing the live-migration
> feature? Such as disk-io, some useless computations (e.g.
> ackermann-function ;-))?

As I said, I was compiling the linux kernel in the domain. (In a loop.) 
Up until the 9762 software, this was quite unreliable. (Resulting in 
panics of the domain and application failures.) So the fact that I was 
able to do this was quite heartening for me.

What is needed now is some kind of stress/reliability test to guarantee 
that everything is actually correct instead of not just obviously 
broken, but I don't know what a good test will be off-hand.

> 
> 
> Thanks,
> Martin
>