[ovs-dev] [PATCH] ovn-controller: Handle physical changes correctly
Liran Schour
LIRANS at il.ibm.com
Mon Jul 25 07:44:42 UTC 2016
> From: Ryan Moats <rmoats at us.ibm.com>
> To: dev at openvswitch.org
> Date: 23/07/2016 12:57 AM
> Subject: [ovs-dev] [PATCH] ovn-controller: Handle physical changes
correctly
> Sent by: "dev" <dev-bounces at openvswitch.org>
>
> [1] reported increased failure rates in certain tests
> with incremental processing (the numbers are the number of failures
> seen in 100 tests):
>
> 2 ovn -- vtep: 3 HVs, 1 VIFs/HV, 1 GW, 1 LS
> 10 ovn -- 2 HVs, 2 LS, 1 lport/LS, 2 peer LRs
> 52 ovn -- 1 HV, 1 LS, 2 lport/LS, 1 LR
> 45 ovn -- 1 HV, 2 LSs, 1 lport/LS, 1 LR
> 23 ovn -- 2 HVs, 3 LS, 1 lport/LS, 2 peer LRs, static routes
> 53 ovn -- 2 HVs, 3 LRs connected via LS, static routes
> 32 ovn -- 2 HVs, 2 LRs connected via LS, gateway router
> 50 ovn -- icmp_reply: 1 HVs, 2 LSs, 1 lport/LS, 1 LR
>
> These failures were caused by a combination of problems in
> handling physical changes:
>
> 1. When a vif was removed, the localvif_to_ofport entry was not
> removed.
> 2. When a physical change was detected, ovn-controller would wait
> a poll cycle before processing the logical flow table.
>
> This patch set addresses both of these issues while simultaneously
> cleaning up the code in physical.c. A side effect is a modification
> of where OF flows are dumped in the gateway router case that allowed
> the root causes of this issue to be found.
>
> With these changes, all of the above tests had a 100/100 success rate.
>
> [1] http://openvswitch.org/pipermail/dev/2016-July/075803.html
>
> Signed-off-by: Ryan Moats <rmoats at us.ibm.com>
> ---
This patch really improves the tests however on my performance evaluation
environment it results in a 100% CPU load on the ovn-controllers.
More information about the dev
mailing list