On 2/28/25 12:56 PM, Ales Musil wrote:
> On Fri, Jan 31, 2025 at 11:15 AM Xavier Simonart <[email protected]>
> wrote:
>
>> When two chassis claim the same port, if a port is in the MARK_UP
>> state when the other chassis claims it, then ovn-controller was
>> sometimes using 100% cpu, as if-status was fighting to add ovn-install
>> (as in the MARK_UP state) and binding was fighting to remove it (as
>> the port was bound to the other chassis).
>>
>> Signed-off-by: Xavier Simonart <[email protected]>
>>
>> ---
>> v2: - rebase on origin/main for scapy fix.
>> - rebase on origin/main.
>> v3: - rebase on origin/main
>> v4: - rebase on origin/main
>> - sent as separate patch, and not in a serie on top of following patch
>> which
>> needs rework: "controller: Optimize postponed ports handling."
>> ---
[...]
>> diff --git a/tests/ovn.at b/tests/ovn.at
>> index 8ecf1f6bf..67099dda1 100644
>> --- a/tests/ovn.at
>> +++ b/tests/ovn.at
>> @@ -16443,6 +16443,9 @@ max_claims=20
>> AT_CHECK([test "${hv1_claims}" -le "${max_claims}"], [0], [])
>> AT_CHECK([test "${hv2_claims}" -le "${max_claims}"], [0], [])
>>
>> +n_uninstalled=$(as hv1 grep -c 'Removing iface vif ovn-installed in'
>> hv1/ovn-controller.log)
>> +echo "$n_uninstalled times 'Removing iface vif ovn-installed'"
>> +AT_CHECK([test "${n_uninstalled}" -le "10"], [0], [])
Was this why this test was flaky in CI (and why it's tagged with
TAG_UNSTABLE)? If so, should we follow up and remove the unstable tag?
>> check ovn-nbctl --wait=hv lsp-del lsp0
>> CHECK_AFTER_RECOMPUTE([hv1], [hv1])
>>
>> --
>> 2.47.1
>>
>> _______________________________________________
>> dev mailing list
>> [email protected]
>> https://mail.openvswitch.org/mailman/listinfo/ovs-dev
>>
>>
> Looks good to me, thanks.
> Acked-by: Ales Musil <[email protected]>
>
Thanks, Xavier and Ales!
Applied to main, 25.03 and 24.09. It didn't apply cleanly to 24.03.
Do we need a custom backport patch for 24.03?
Regards,
Dumitru
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev