[ovs-discuss] Upgraded to openvswitch-1.4.1 and still high load and polluted syslog
Oliver Francke
Oliver.Francke at filoo.de
Tue May 29 18:01:10 UTC 2012
Hi Ben, Justin,
Am 29.05.2012 um 19:35 schrieb Ben Pfaff:
> On Tue, May 29, 2012 at 09:39:19AM -0700, Justin Pettit wrote:
>> On May 29, 2012, at 5:02 AM, Oliver Francke wrote:
>>
>>> Hi Justin/Volkan,
>>>
>>> On 05/24/2012 06:50 PM, Justin Pettit wrote:
>>>> On May 24, 2012, at 8:43 AM, Volkan YAZICI wrote:
>>>>
>>>>> I always thought STP would have been enabled by default.
>>>>
>>>> vswitches are typically deployed at the edge, so there isn't usually a need for STP in those deployments. STP support is a relatively recent addition, and since OVS is usually run as a vswitch, we decided to leave it off by default.
>>>
>>> well... that's not the worst decision. If I add/delete tap-interfaces for the qemu-kvm processes, there is a connectivity-timeout on the whole machine, which I now refer to an STP-relevant event.
>>> How can I make sure, that interface-up/down does _not_ influence the ovs-stp-functionality, if at all possible?
>>
>> What if you disable STP on those tap devices? You can control this
>> with the "other_config : stp-enable" key on the Port table. I
>> imagine you'll want to disable STP in the same atomic transaction.
>> You can do this by using the "--" separator in a single ovs-vsctl
>> command to add the port and disable STP.
>
> Or even:
> ovs-vsctl add-port br0 portname other-config:stp-enable=false
yes, thank you very much, have been there already, but was misleaded by the bridge-command-syntax from the man-page:
ovs-vsctl set Bridge br0 stp_enable=true
so I thought it would be… "other_config:stp_enable=…" with hyphen.
Next change-window will be tomorrow, looking forward now.
Thank you very much and best regards,
Oliver.
> _______________________________________________
> discuss mailing list
> discuss at openvswitch.org
> http://openvswitch.org/mailman/listinfo/discuss
More information about the discuss
mailing list