On Thu, Sep 10, 2020 at 11:04 AM Ilya Maximets <[email protected]> wrote:
> On 9/10/20 7:05 PM, Han Zhou wrote: > > On Thu, Sep 10, 2020 at 8:39 AM Numan Siddique <[email protected]> wrote: > >> > >> On Wed, Sep 9, 2020 at 1:41 AM Han Zhou <[email protected]> wrote: > >>> > >>> This command is added to avoid the flooding of nb_cfg updates from > >>> a large nubmer of hypervisors interfering with the original SB DB > >>> data distribution and handling, so that the e2e control plane latency > >>> can be measured more accurately during scale testing with the command > >>> ovn-nbctl --print-wait-time --wait=hv ... > >>> > >>> Without this, there can be hypervisors updating back to SB before the > >>> SB server finishes sending notifications to the rest of hypervisors. > >>> In my test with 3K HVs, delaying the report for 2 seconds is enough to > >>> get accurate latency report. > >>> > >>> Signed-off-by: Han Zhou <[email protected]> > >> > >> Acked-by: Numan Siddique <[email protected]> > >> > >> Note: You have to rebase before applying. > >> > > Thanks Numan. I applied this to master. > > The test in this patch is not stable and fails on travis from time to time: > https://travis-ci.org/github/ovn-org/ovn/jobs/726006756#L5318 > > Probably, 3 seconds is not enough, or maybe there is a real bug somewhere. > > Best regards, Ilya Maximets. > Thanks Ilya, I just sent a fix: https://patchwork.ozlabs.org/project/ovn/patch/[email protected]/ Could you take a look? Sorry for the inconvenience. Han _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
