Re: [ovs-discuss] encapsulation and MTUs (was: Re: Path MTU Discovery)

2014-02-03 Thread Jesse Gross
On Fri, Jan 31, 2014 at 11:20 AM, David Magda wrote: > On Thu, January 30, 2014 15:45, Jesse Gross wrote: >>> A particular detail: on which host (i.e, non-VM) interfaces does one >>> change the MTU for OpenStack? For example, on OS compute nodes, there is >>> the actual NIC (eth0), but also br-int

Re: [ovs-discuss] encapsulation and MTUs (was: Re: Path MTU Discovery)

2014-01-31 Thread David Magda
On Thu, January 30, 2014 15:45, Jesse Gross wrote: >> A particular detail: on which host (i.e, non-VM) interfaces does one >> change the MTU for OpenStack? For example, on OS compute nodes, there is >> the actual NIC (eth0), but also br-int and br-tun (where the GRE tunnels >> connect to): > > You

Re: [ovs-discuss] encapsulation and MTUs (was: Re: Path MTU Discovery)

2014-01-30 Thread Jesse Gross
On Wed, Jan 29, 2014 at 7:26 AM, David Magda wrote: > On Wed, January 29, 2014 01:11, Thomas Glanzmann wrote: > >> In order to avoide the problem you either have to lower the MTU in the >> VMs which you did or you enable bigger packets in your transport >> network. This is called jumbo frames. On

[ovs-discuss] encapsulation and MTUs (was: Re: Path MTU Discovery)

2014-01-29 Thread David Magda
On Wed, January 29, 2014 01:11, Thomas Glanzmann wrote: > In order to avoide the problem you either have to lower the MTU in the > VMs which you did or you enable bigger packets in your transport > network. This is called jumbo frames. On HP procurve switches they're > enabled by saying A particu