| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN] Change oslo config options entry point for the OVN agent https://review.opendev.org/c/openstack/neutron/+/876824 | 08:47 |
|---|---|---|
| opendevreview | Rodolfo Alonso proposed openstack/neutron stable/2023.1: [OVN] Change oslo config options entry point for the OVN agent https://review.opendev.org/c/openstack/neutron/+/876825 | 08:47 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron-tempest-plugin master: Move test_dhcp_port_status_active to tempest https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/869227 | 09:31 |
| ralonsoh | lajoskatona, slaweq he folks, https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/869227 is not getting into zuul (I'm asking in the infra channel) | 09:34 |
| ralonsoh | in any case, is that 100% necessary for the release? | 09:34 |
| ralonsoh | https://review.opendev.org/c/openstack/releases/+/876616 | 09:34 |
| ralonsoh | or can I update it now? | 09:35 |
| lajoskatona | ralonsoh: thanks for checking, It is not that urgent, it is not that we lose coverage without it, we have no duplication | 10:02 |
| lajoskatona | ralonsoh: as I see now it is in the gate queue | 10:02 |
| ralonsoh | it is now?? | 10:03 |
| ralonsoh | It has appeared now, I swear | 10:03 |
| lajoskatona | ralonsoh: ohh, no it is in the release-approval queue | 10:03 |
| ralonsoh | I +W it 10 mins ago | 10:03 |
| ralonsoh | ok, I'll wait for it | 10:04 |
| opendevreview | Slawek Kaplonski proposed openstack/neutron stable/zed: [Secure RBAC] Add shared_security_groups rule https://review.opendev.org/c/openstack/neutron/+/876802 | 10:08 |
| opendevreview | Slawek Kaplonski proposed openstack/neutron-tempest-plugin master: [Secure RBAC] Add scope enforcement enabled job for Zed branch https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/874709 | 10:08 |
| slaweq | ralonsoh hi, I'm still checking this Zed job with new policies but for sure we need new neutron-lib release for Zed: https://review.opendev.org/c/openstack/releases/+/876837 | 10:48 |
| slaweq | as we need https://review.opendev.org/c/openstack/neutron-lib/+/874394 there | 10:48 |
| opendevreview | Slawek Kaplonski proposed openstack/neutron-tempest-plugin master: [Secure RBAC] Add scope enforcement enabled job for Zed branch https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/874709 | 10:50 |
| ralonsoh | slaweq, no no, I've pushed these patches today | 10:51 |
| ralonsoh | https://review.opendev.org/q/project:openstack%252Freleases | 10:51 |
| ralonsoh | check the mail | 10:51 |
| slaweq | ahh, ok | 10:51 |
| slaweq | so please -1 this one | 10:51 |
| opendevreview | Slawek Kaplonski proposed openstack/neutron-tempest-plugin master: [Secure RBAC] Add scope enforcement enabled job for Zed branch https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/874709 | 10:53 |
| slaweq | ralonsoh++ thx | 10:53 |
| ralonsoh | yw! | 10:53 |
| slaweq | ok, so for now we for sure need that one missing backport which I proposed today and You +2 already and new neutron-lib release | 10:55 |
| slaweq | then I will check again how results will look like | 10:55 |
| slaweq | but I don't see anything else missing comparing it to master branch | 10:56 |
| ralonsoh | once we have the new n-lib release, we'll wait for the requirements patch | 10:56 |
| ralonsoh | and then we'll be able to test n-t-p patch | 10:56 |
| slaweq | ++ | 10:56 |
| ralonsoh | including this last backport | 10:56 |
| opendevreview | Sahid Orentino Ferdjaoui proposed openstack/neutron stable/2023.1: ovs: fix regression when vlan mapping is not already registered https://review.opendev.org/c/openstack/neutron/+/876803 | 10:59 |
| opendevreview | Sahid Orentino Ferdjaoui proposed openstack/neutron stable/zed: ovs: fix regression when vlan mapping is not already registered https://review.opendev.org/c/openstack/neutron/+/876804 | 11:00 |
| opendevreview | Bence Romsics proposed openstack/neutron master: DNM Suppress IPv6 metadata DAD failure https://review.opendev.org/c/openstack/neutron/+/876566 | 11:25 |
| opendevreview | Felix Huettner proposed openstack/neutron master: Reduce lock contention on subnets https://review.opendev.org/c/openstack/neutron/+/875938 | 12:56 |
| *** elodilles is now known as elodilles_afk | 13:20 | |
| opendevreview | Slawek Kaplonski proposed openstack/neutron-tempest-plugin master: DNM Just test of the Zed job with latest stable/zed neutron-lib https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/876869 | 13:25 |
| opendevreview | Slawek Kaplonski proposed openstack/neutron-tempest-plugin master: DNM Just test of the Zed job with latest stable/zed neutron-lib https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/876869 | 13:26 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN] Use the BW values retrieved from ``get_port_qos`` https://review.opendev.org/c/openstack/neutron/+/876872 | 13:53 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron stable/2023.1: [OVN] Use the BW values retrieved from ``get_port_qos`` https://review.opendev.org/c/openstack/neutron/+/876806 | 13:54 |
| *** elodilles_afk is now known as elodilles | 14:58 | |
| opendevreview | Bence Romsics proposed openstack/neutron master: Suppress IPv6 metadata DAD failure https://review.opendev.org/c/openstack/neutron/+/876566 | 15:04 |
| haleyb | rubasov: hey bence, i think i figured out a way to fix the IPv6 metadata issue, just came to me | 15:05 |
| rubasov | haleyb: hi, tell me | 15:06 |
| rubasov | (by the way I also just pushed a new patch, but did not write an explanation for it yet in gerrit) | 15:06 |
| haleyb | so looking at this i noticed that EC2 also supports this now, but did not use a link-local, they used a ULA address, fd00:ec2::254/128 | 15:07 |
| haleyb | i'll type quick, have more comments, but will put in bug as well... | 15:07 |
| haleyb | if we configure that on the loopback device and inject a route to it in the dhcp reply, it should work | 15:08 |
| haleyb | we do this for the IPv4 case already | 15:08 |
| haleyb | i had actually started hacking something to support multiple IPv6 metadata addresses yesterday, and when i got to the route injection part i said 'a-ha!' | 15:09 |
| haleyb | did that make sense? | 15:09 |
| haleyb | it would mean we have to change the metadata address, since i couldn't put the a9fe address on lo and have it work | 15:09 |
| haleyb | i could get a PoC patch out today that does the whole mess i think, i might have gone overboard but was a good thought experiment | 15:11 |
| rubasov | then I guess each guest would have the route corresponding to where they got the dhcp offer... | 15:12 |
| haleyb | rubasov: exactly, just like IPv4 - see code in _generate_opts_per_subnet() | 15:12 |
| rubasov | what would happen when that dhcp server fails and dhcp has to go to the other server which has a different route to push? | 15:12 |
| haleyb | i would hope the VM would update the route, but it shouldn't need to do metadata again, right? | 15:13 |
| haleyb | i guess technically we should have been doing this for IPv4 as well, configuring on the loopback, it just never does DAD | 15:15 |
| rubasov | interesting idea, I will play with it to understand it better | 15:17 |
| haleyb | let me paste the two commands | 15:18 |
| ralonsoh | in order to make this change, that is disruptive with the current behaviour, we should discuss this in the drivers meeting | 15:18 |
| haleyb | on dhcp agent: ip a a dev lo fd00:ec2::254/128 | 15:18 |
| haleyb | on VM: ip -6 r a fd00:ec2::254/128 via DHCP-LL dev MYDEV | 15:19 |
| haleyb | ralonsoh: yes, it is disruptive, but we could have them live together until we update cloud-init. it is just a thought | 15:20 |
| ralonsoh | why this IPv6 address? | 15:20 |
| haleyb | ralonsoh: well, i just chose the EC2 one, and it's ULA, not link-local, so we can route to it | 15:20 |
| rubasov | haleyb: thanks for the commands | 15:21 |
| rubasov | no router will forward the LL address | 15:21 |
| ralonsoh | though what interface? the DHCP agent external interface? | 15:21 |
| haleyb | ralonsoh: it's internal interface. Look for METADATA_CIDR in _generate_opts_per_subnet() - we do it today for v4 | 15:22 |
| ralonsoh | the DHCP namespace has two interfaces, the one connected to OVS and lo | 15:23 |
| ralonsoh | where is this IPv6 set? | 15:23 |
| rubasov | in a guest on an isolated network I have a routing table like this: | 15:23 |
| rubasov | $ ip r | 15:23 |
| rubasov | default via 10.0.4.1 dev eth0 | 15:23 |
| rubasov | 10.0.4.0/24 dev eth0 scope link src 10.0.4.76 | 15:23 |
| rubasov | 169.254.169.254 via 10.0.4.2 dev eth0 | 15:23 |
| rubasov | 169.254.169.254 via 10.0.4.3 dev eth0 | 15:23 |
| haleyb | ralonsoh: my thought it lo, since i think putting on eth0 would just trigger DAD | 15:23 |
| haleyb | rubasov: oh, it has both dhcp agent IPs, interesting | 15:24 |
| rubasov | haleyb: I'm also surprised | 15:24 |
| haleyb | thrice actually | 15:24 |
| haleyb | oh, guess not, the .1 is the gateway | 15:25 |
| ralonsoh | what IPv6 address will have the one connected to OVS? | 15:25 |
| rubasov | one agent pushes this: tag:subnet-9a15f085-a202-44a0-9d97-502f8752ca85,option:classless-static-route,169.254.169.254/32,10.0.4.3,0.0.0.0/0,10.0.4.1 | 15:25 |
| rubasov | the other pushes this: tag:subnet-9a15f085-a202-44a0-9d97-502f8752ca85,option:classless-static-route,169.254.169.254/32,10.0.4.2,0.0.0.0/0,10.0.4.1 | 15:26 |
| haleyb | rubasov: did you fail-over from one to the other? i guess in that case it could have added the second, since it's not technically a duplicate | 15:27 |
| haleyb | but like i said, metadata is long done | 15:27 |
| rubasov | (and IIRC option 249 is some microsoft counterpart to classless-static-route) | 15:27 |
| rubasov | haleyb: I did a lot of things with this test environment already, I would not start with a clean state to make sure | 15:28 |
| rubasov | * I would need | 15:28 |
| haleyb | rubasov: ack, like i said, i'll try and get something out today. it is adding the addresses on 'eth0' now so would just have to fix that somehow | 15:29 |
| rubasov | okay, waiting for it | 15:29 |
| rubasov | in the short term, this still may make sense to get rid of the dhcp side effects of the uncaught exception: https://review.opendev.org/c/openstack/neutron/+/876566 | 15:30 |
| * haleyb has a few meetings, but will get something out at least | 15:30 | |
| rubasov | removed the DNM, because I realized that it only improves things, even if it does not fix everything it should | 15:31 |
| ralonsoh | ok, I think I'm out the conversation | 15:31 |
| opendevreview | Merged openstack/neutron-tempest-plugin master: Move test_dhcp_port_status_active to tempest https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/869227 | 15:40 |
| ralonsoh | slaweq, please approve this patch | 15:43 |
| ralonsoh | https://review.opendev.org/c/openstack/neutron/+/876804 | 15:43 |
| ralonsoh | to add it to the next Zed release | 15:43 |
| ralonsoh | lajoskatona, ^ | 15:54 |
| ralonsoh | (just to update the release patch asap) | 15:54 |
| slaweq | ralonsoh done | 15:58 |
| slaweq | ralonsoh I tested zed new rbac job in https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/876869 with neutron-lib from stable/zed branch and it is green | 15:59 |
| slaweq | so once we will have that last patch merged in neutron and neutron-lib released we should be good with it :) | 15:59 |
| ralonsoh | thanks! | 16:03 |
| ralonsoh | slaweq, with n-lib master cool! | 16:03 |
| ralonsoh | now we have the confirmation before releasing | 16:03 |
| ralonsoh | ++ | 16:03 |
| slaweq | ralonsoh to be strict, not n-lib master but n-lib stable/zed's head :) | 16:04 |
| ralonsoh | yeah hehehe | 16:04 |
| slaweq | but yes, we have confirmation that it works with it | 16:04 |
| opendevreview | Merged openstack/neutron stable/zed: [Secure RBAC] Add shared_security_groups rule https://review.opendev.org/c/openstack/neutron/+/876802 | 16:10 |
| opendevreview | Merged openstack/neutron stable/2023.1: ovs: fix regression when vlan mapping is not already registered https://review.opendev.org/c/openstack/neutron/+/876803 | 16:10 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVS] Allow custom ethertype traffic in the ingress table https://review.opendev.org/c/openstack/neutron/+/876563 | 17:51 |
| haleyb | ralonsoh: so you will fix the other bug i noticed in https://review.opendev.org/c/openstack/neutron/+/876872 separately? if so i can change my vote | 18:30 |
| opendevreview | Merged openstack/neutron master: [sqlalchemy-20] The Session.begin.subtransactions flag is deprecated https://review.opendev.org/c/openstack/neutron/+/874938 | 20:07 |
| opendevreview | Merged openstack/neutron stable/wallaby: Prevent router_ha_interface port from being removed via API https://review.opendev.org/c/openstack/neutron/+/875820 | 21:25 |
| opendevreview | Brian Haley proposed openstack/neutron master: Add support for multiple IPv6 metadata addresses https://review.opendev.org/c/openstack/neutron/+/876903 | 23:16 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!