Thursday, 2022-01-13

*** kmasterson` is now known as kmasterson07:11
*** tbachman_ is now known as tbachman07:31
opendevreviewTakashi Kajinami proposed openstack/neutron master: ml2 ovs: Deprecate unused [agent] veth_mtu  https://review.opendev.org/c/openstack/neutron/+/82444607:57
opendevreviewSlawek Kaplonski proposed openstack/neutron master: Register cleanup hooks for the object change handlers only when needed  https://review.opendev.org/c/openstack/neutron/+/82450808:08
slaweqralonsoh: hi, ^^ it seems that this helped for those FT in stable/train (we talked about it yesterday)08:10
slaweqso I proposed such change in master, I don't think it can really break anything in the production code :)08:11
slaweqlajoskatona: if You will have some time, please take a look at it too :)08:11
slaweqthx in advance08:11
lajoskatonaslaweq: sure, I will check08:25
opendevreviewTakashi Kajinami proposed openstack/neutron master: ml2 ovs: Deprecate unused [agent] veth_mtu  https://review.opendev.org/c/openstack/neutron/+/82444608:30
ralonsohslaweq, sure, let me check08:31
*** mnasiadka_ is now known as mnasiadka08:43
opendevreviewLajos Katona proposed openstack/neutron-tempest-plugin master: QoS min pps API tests  https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/80668908:56
opendevreviewTakashi Kajinami proposed openstack/neutron master: ml2 ovs: Deprecate unused [agent] veth_mtu  https://review.opendev.org/c/openstack/neutron/+/82444608:59
opendevreviewTakashi Kajinami proposed openstack/neutron master: ml2 ovs: Deprecate unused [agent] veth_mtu  https://review.opendev.org/c/openstack/neutron/+/82444609:00
*** ykarel is now known as ykarel|away09:47
mnasiadkaHi there - I'd like to discuss a bug related to ML2/OVN and Magnum's usage of allowed_address_pairs. Bug: https://bugs.launchpad.net/neutron/+bug/1908382 description of the situation: https://paste.openstack.org/show/812087/10:14
mnasiadkaTL;DR - Magnum is spawning instances and sets allowed_address_pairs to /16 CIDR - although it works for ML2/OVS - it doesn't for ML2/OVN10:15
ralonsohcore OVN does not allow to use CIDR in the NAT "dnat_and_snat" registers10:15
ralonsohyou can use /32 IP addresses10:15
ralonsohhttps://bugzilla.redhat.com/show_bug.cgi?id=198633710:15
ralonsohthere you have (now public) the description of the issue10:15
mnasiadkaWhile I was able to work it around, by adding security group rules on the instance ports to accept the additional CIDR network traffic - I don't think it's a good permanent solution ;-)10:17
ralonsohthat could work for internal IP addresses, but the allowed address are added to the FIP address10:18
ralonsohat least for FIP VIPs10:19
mnasiadkaRight, so even if the traffic is internal (between instances in OpenStack) - the mentioned issue in bugzilla is applicable, because the primary IP addresses on that port have a FIP configured?10:20
ralonsohwhat I mean is that the SG rules apply to the internal port IP, not the FIP10:21
ralonsohthe SG rules won't affect the FIP traffic10:21
ralonsohin other words, those SG rules won't filter/allow any traffic for the FIP10:22
mnasiadkaSure, but this traffic (the 10.100.0.0/16 CIDR) is only for internal traffic between Kubernetes nodes.10:22
mnasiadkathe port has two addresses (the one allocated by Nova + 10.100.0.0/16 CIDR used by the Kubernetes used overlay - Calico or Flannel)10:23
ralonsohI don't see (I should check it) how the SG rules create the corresponding NAT registers in OVN that allow this traffic10:25
ralonsohin any case, this is a workaround, as you commented10:25
mnasiadkaYes, what is interesting - ICMP traffic between addresses in 10.100.0.0/16 works without the additional SG rules - but TCP/UDP doesn't.10:28
mnasiadkahttps://review.opendev.org/c/openstack/magnum/+/773923/3/magnum/drivers/k8s_fedora_coreos_v1/templates/kubecluster.yaml - here is the patch to Magnum10:28
mnasiadkaQuestion if this would affect anything for ML2/OVS deployments.10:31
opendevreviewLajos Katona proposed openstack/networking-odl master: Fix some requirements  https://review.opendev.org/c/openstack/networking-odl/+/82342810:31
opendevreviewLajos Katona proposed openstack/networking-odl master: Fix some requirements  https://review.opendev.org/c/openstack/networking-odl/+/82342810:51
*** dasm|off is now known as dasm12:08
opendevreviewMerged openstack/neutron master: ml2 ovs: Deprecate unused [agent] veth_mtu  https://review.opendev.org/c/openstack/neutron/+/82444612:11
holtgrew1Hello. I have setup double-homed OpenStack hosts based on VLANs. They are on one physical port. The default VLAN is "671" and goes into bridge brbond0 that contains the bond bond0. The other VLAN is 678 and brpublic contains brbond0.678 (Rocky/CentOS interface naming). I'm seeing ICMPv6 traffic from VLAN 678 in bond0. How can I find out whether that data is traveling through neutron or is 12:42
holtgrew1somehow forwarded on the host?12:42
opendevreviewLajos Katona proposed openstack/neutron-vpnaas master: Change BUILD_OVS_FROM_SOURCE to Q_BUILD_OVS_FROM_GIT  https://review.opendev.org/c/openstack/neutron-vpnaas/+/82456012:49
fricklerholtgrew1: if I understand your setup correctly, that would be the expected behavior, you also see the tagged traffic on the base interface. "tcpdump -e" should show you the vlan headers to verify12:54
opendevreviewMerged openstack/neutron master: [Functional] Add extra logs to the L3 HA router transitions  https://review.opendev.org/c/openstack/neutron/+/82409812:54
opendevreviewKrzysztof Tomaszewski proposed openstack/neutron master: Exclude router gw subnet ports from port list while port delete.  https://review.opendev.org/c/openstack/neutron/+/82400812:55
lajoskatonaslaweq, ralonsoh: if you have time please check https://review.opendev.org/c/openstack/os-ken/+/824476 , neutron-tempest-plugin-dynamic-routing job fails due to my mistake (https://review.opendev.org/c/openstack/os-ken/+/824215 )13:29
ralonsohsure13:29
lajoskatonaslaweq, ralonsoh: in 824476  I added neutron-tempest-plugin-dynamic-routing job to os-ken, as there is no unit test coverage for bgp (and many other things) in os-ken13:30
lajoskatonaPerhaps worth to keep it for check queue at least13:30
ralonsohI though this call has always password in kwargs13:30
lajoskatonaralonsoh: this is why dicts are a way to hell :-) objects to everywhere (it was a joke only)....13:31
lajoskatonaI thought the same and in my old local devstack with n-d-r I had it.13:32
lajoskatonaall: tomorrow I will have a whole day workshop, so I will be less active (I will join the drivers meeting, and will send out the mail about the agenda this afternoon as usual)13:36
slaweqlajoskatona: done13:48
opendevreviewMerged openstack/neutron master: Remove "PortBindingMixin" class and related DB table  https://review.opendev.org/c/openstack/neutron/+/82400914:11
opendevreviewLajos Katona proposed openstack/os-ken stable/xena: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454614:58
opendevreviewMerged openstack/os-ken master: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82447615:04
opendevreviewLajos Katona proposed openstack/os-ken stable/wallaby: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454715:11
opendevreviewLajos Katona proposed openstack/os-ken stable/victoria: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454815:11
opendevreviewMamatisa Nurmatov proposed openstack/neutron-tempest-plugin master: Add Local IP API tests  https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/81474215:21
opendevreviewMamatisa Nurmatov proposed openstack/neutron-tempest-plugin master: Add local ip scenario tests  https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/82300715:21
opendevreviewLajos Katona proposed openstack/os-ken stable/victoria: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454815:21
opendevreviewLajos Katona proposed openstack/os-ken stable/ussuri: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454915:22
opendevreviewMerged openstack/neutron master: Add Local IP policy rules  https://review.opendev.org/c/openstack/neutron/+/81643515:22
opendevreviewMerged openstack/neutron stable/xena: [OVN] Allow only one physical network per bridge  https://review.opendev.org/c/openstack/neutron/+/82434415:22
opendevreviewLajos Katona proposed openstack/os-ken stable/train: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82461015:23
opendevreviewLajos Katona proposed openstack/os-ken stable/xena: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454615:26
opendevreviewLajos Katona proposed openstack/os-ken stable/xena: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454615:36
opendevreviewLajos Katona proposed openstack/os-ken stable/wallaby: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454715:37
opendevreviewMerged openstack/neutron master: [OVN] Check if OVN NB supports stateless NAT rules  https://review.opendev.org/c/openstack/neutron/+/81637615:38
opendevreviewLajos Katona proposed openstack/os-ken stable/victoria: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454815:39
opendevreviewLajos Katona proposed openstack/os-ken stable/ussuri: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82454915:40
opendevreviewLajos Katona proposed openstack/os-ken stable/train: Avoid missing key 'password' for neighbor_add  https://review.opendev.org/c/openstack/os-ken/+/82461015:40
opendevreviewRodolfo Alonso proposed openstack/neutron master: [OVN] Check if OVN SB supports virtual ports  https://review.opendev.org/c/openstack/neutron/+/81638316:04
ralonsohslaweq, ^^ that was just a rebase, if you don't mind16:05
slaweqralonsoh: done16:05
ralonsohthanks!16:06
ralonsohslaweq, I'm going to push https://review.opendev.org/c/openstack/neutron/+/824508 on top of https://review.opendev.org/c/openstack/neutron/+/82252216:11
ralonsohin Train16:11
ralonsohjust to check the CI16:11
slaweqralonsoh: I did something like that yesterday https://review.opendev.org/c/openstack/neutron/+/824450/116:12
ralonsohahh ok16:12
slaweqit wasn't exactly the same as e.g. name of function was different16:12
slaweqbut in general it was that patch done for stable/train to test it there16:12
ralonsohso that's what I was looking for16:12
slaweqand it helped for functional tests, both py36 and py2716:12
ralonsohthis is passing now16:12
ralonsohso my +2 for this patch16:12
ralonsohbecause this is actually solving Train, where you found the error16:13
ralonsohsolving/fixing16:13
slaweqthx a lot16:15
opendevreviewBalazs Gibizer proposed openstack/neutron master: DNM: trigger tempest  https://review.opendev.org/c/openstack/neutron/+/82460516:39
lajoskatonaralonsoh: could you please check this doc change when you have few minutes: https://review.opendev.org/c/openstack/neutron/+/809579 , it is just changing when updating QoS Policy  with min_bw is available (not Victoria but Wallaby)16:49
ralonsohlajoskatona, sure16:49
opendevreviewRodolfo Alonso proposed openstack/neutron master: Create an index for "agents.host" column  https://review.opendev.org/c/openstack/neutron/+/82382217:11
lajoskatonaralonsoh: thanks17:15
opendevreviewMerged openstack/neutron master: doc: Change availability of QoS policy change  https://review.opendev.org/c/openstack/neutron/+/80957917:44
*** marlinc is now known as Guest49118:00
*** beagles_ is now known as eagles18:48
opendevreviewSlawek Kaplonski proposed openstack/neutron master: Add upgrade check for extra DHCP options  https://review.opendev.org/c/openstack/neutron/+/82464119:40
*** dasm is now known as dasm|off21:45

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!