[ovs-dev] [PATCH v2 3/3] secchan: Better tolerate failing controller admission control in fail-open.

Peter Balland peter at nicira.com
Wed Sep 16 22:06:04 UTC 2009


Ben Pfaff wrote:
> Pete, you're completely right about this.  I think that we can do
> better than what this commit does, if we do some work on the NOX
> side.
> 
> What I really want to do right now is, then, this:
> 
>         - Push this as-is.
> 
>         - File a bug describing my thoughts on how to fix it
>           properly..
> 
>         - Fix that bug later, either just in NOX or in OpenFlow
>           as a whole if we can get the consortium to agree.
> 
> Does that make sense, or do you think it should be fixed all at
> once?
> 
> (This is semi-urgent, because the current implementation of
> fail-open appears to sometimes causes HA failures and thus host
> reboots on XenServer.)

I agree that a fix is urgently needed for this, and don't see any 
problems from pushing the fix as is, so long as it is only as a stopgap 
measure.  I think we should push hard on getting a more permanent 
solution in place ASAP, so I'll take this as an action item to make sure 
the appropriate requirements get communicated to to the OpenFlow consortium.

Peter




More information about the dev mailing list