| *** mnasiadka_ is now known as mnasiadka | 06:29 | |
| *** elodilles_pto is now known as elodilles | 07:41 | |
| ralonsoh | amotoki, hi, can you check https://review.opendev.org/c/openstack/releases/+/862911? | 08:02 |
|---|---|---|
| ralonsoh | thanks in advance | 08:03 |
| amotoki | ralonsoh: sure | 08:05 |
| labedz__ | hi guys: could you take a look on those backports: https://review.opendev.org/c/openstack/neutron/+/863073 , https://review.opendev.org/c/openstack/neutron/+/863074 | 08:44 |
| labedz__ | thx! | 08:44 |
| *** labedz__ is now known as labedz | 08:44 | |
| ralonsoh | ok | 08:50 |
| ralonsoh | folks, a heads-up | 09:16 |
| ralonsoh | EOL transition for Q, R and S | 09:16 |
| ralonsoh | https://review.opendev.org/c/openstack/releases/+/862937 | 09:16 |
| bbezak | ralonsoh mnasiadka https://bugs.launchpad.net/neutron/+bug/1995287, here you have bug opened for dhcp-agent issue with baremetal on yoga | 09:43 |
| ralonsoh | bbezak, but the problem you have is that your DHCP is requesting a non existing network | 09:56 |
| ralonsoh | what is the server log when the DHCP requests that network? | 09:57 |
| bbezak | ralonsoh - network in fact do exist. let me see those errors | 10:00 |
| ralonsoh | bbezak, do you have the logs of the DHCP agent in debug mode? | 10:00 |
| bbezak | ralonsoh: mnasiadka pasted here - https://bugs.launchpad.net/neutron/+bug/1995287 | 10:01 |
| ralonsoh | bbezak, I only see INFO messages | 10:02 |
| mnasiadka | updated with DEBUG log from both dhcp agent and neutron-server.log | 10:03 |
| ralonsoh | mnasiadka, where? | 10:03 |
| mnasiadka | In the ticket | 10:04 |
| ralonsoh | I see no DEBUG messages anywhere | 10:04 |
| mnasiadka | https://bugs.launchpad.net/neutron/+bug/1995287/comments/1 - that's not DEBUG? | 10:04 |
| ralonsoh | mnasiadka, ok, let's start again. In order to debug this issue, it is needed the full DHCP agent and server logs, both in debug mode | 10:05 |
| ralonsoh | if there is a missing network, according to the DHCP agent, then we'll need to "find" it | 10:05 |
| ralonsoh | if this network is present, then we'll need to check what information we are passing to the DHCP agent from the server | 10:06 |
| ralonsoh | you said you bumped the version | 10:06 |
| ralonsoh | did you bump the Neutron server and the DHCP agent? | 10:06 |
| ralonsoh | both services | 10:06 |
| mnasiadka | yes, restarted multiple times | 10:09 |
| mnasiadka | uploaded files to the bug, it's a lot of megabytes | 10:09 |
| ralonsoh | mnasiadka, what you need to manually check is the content of the Neutron server | 10:09 |
| ralonsoh | the reply | 10:09 |
| ralonsoh | when the DHCP agent requests the network information | 10:10 |
| ralonsoh | mnasiadka, again, did you bump the DHCP agent? | 10:12 |
| ralonsoh | according to the logs | 10:12 |
| ralonsoh | 2022-11-02 09:43:15.664 7 INFO neutron.common.config [-] /var/lib/kolla/venv/bin/neutron-dhcp-agent version 20.2.1.dev53 | 10:12 |
| ralonsoh | this is Yoga | 10:12 |
| ralonsoh | ah, no, from X to Y, that's ok | 10:13 |
| bbezak | yes, we've upgraded to yoga ralonsoh | 10:13 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [WIP] Placement tunnelled networks https://review.opendev.org/c/openstack/neutron/+/860639 | 10:38 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron-specs master: Strict minimum bandwidth support for tunnelled networks https://review.opendev.org/c/openstack/neutron-specs/+/860859 | 10:53 |
| opendevreview | Claudiu Belu proposed openstack/neutron stable/zed: Check subnet overlapping after add router interface https://review.opendev.org/c/openstack/neutron/+/863271 | 11:07 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [WIP][POC] NFTables https://review.opendev.org/c/openstack/neutron/+/759874 | 11:09 |
| ralonsoh | slaweq, hi! | 11:13 |
| ralonsoh | https://review.opendev.org/c/openstack/neutron-specs/+/857858/1..2/specs/2023.1/configurable-default-sg-rules.rst | 11:13 |
| ralonsoh | I'm still missing some comments | 11:13 |
| ralonsoh | L233 | 11:13 |
| ralonsoh | and https://review.opendev.org/c/openstack/neutron-specs/+/857858/1..2/specs/2023.1/configurable-default-sg-rules.rst#166 | 11:14 |
| ralonsoh | the last one should be specified in the spec | 11:14 |
| opendevreview | Adrian Alexandrescu proposed openstack/neutron stable/zed: [OVN] Allow only one physical network per bridge https://review.opendev.org/c/openstack/neutron/+/863255 | 11:36 |
| opendevreview | Adrian Alexandrescu proposed openstack/neutron stable/zed: [OVN] Allow only one physical network per bridge https://review.opendev.org/c/openstack/neutron/+/863256 | 11:37 |
| opendevreview | Stefan Emanuel Boldeanu proposed openstack/neutron stable/yoga: Check subnet overlapping after add router interface https://review.opendev.org/c/openstack/neutron/+/863275 | 11:51 |
| opendevreview | Paval Mihaela-Irina proposed openstack/neutron stable/yoga: Port provisioning should retry only for VM ports https://review.opendev.org/c/openstack/neutron/+/863279 | 11:53 |
| opendevreview | Patache Cosmin - Andrei proposed openstack/neutron stable/zed: Since OVN 20.06, config is stored in "Chassis.other_config" https://review.opendev.org/c/openstack/neutron/+/863280 | 11:57 |
| opendevreview | Patache Cosmin - Andrei proposed openstack/neutron stable/yoga: Since OVN 20.06, config is stored in "Chassis.other_config" https://review.opendev.org/c/openstack/neutron/+/863281 | 11:59 |
| *** elodilles is now known as elodilles_afk | 12:44 | |
| *** gthiemon1e is now known as gthiemonge | 13:04 | |
| *** elodilles_afk is now known as elodilles | 13:15 | |
| opendevreview | Merged openstack/networking-ovn stable/train: Ensure OVN-LB is properly configured upon LS removal from LR https://review.opendev.org/c/openstack/networking-ovn/+/862363 | 13:17 |
| sahid | lajoskatona: o/ I checked around tests and functional tests that we have, but did not find anything that I can use to help me building a full multisgements test-case, do you have any pointers for me that can help me? | 13:24 |
| opendevreview | Merged openstack/neutron stable/ussuri: Revert "[OVN] Set NB/SB "connection" inactivity probe" https://review.opendev.org/c/openstack/neutron/+/863074 | 13:25 |
| lajoskatona | sahid: Hi | 13:26 |
| lajoskatona | sahid: let me check | 13:28 |
| sahid | cool thank you | 13:31 |
| lajoskatona | sahid: for fullstack the framework allows you to add tests where you start neutron-server and agents with their config and other specific things, so that is good, what is missing that yo don't have placement | 13:33 |
| opendevreview | Merged openstack/os-vif master: Move mtu update request into ovsdb transaction https://review.opendev.org/c/openstack/os-vif/+/857868 | 13:35 |
| lajoskatona | sahid: you can use a fake placement, like we did for min_bw tests, but it can be tricky, so check it, (https://opendev.org/openstack/neutron/src/branch/master/neutron/tests/fullstack/servers/placement.py ), but I am not sure if it can be used for multisegments | 13:36 |
| ralonsoh | sean-k-mooney, https://review.opendev.org/c/openstack/os-vif/+/857868 | 13:37 |
| ralonsoh | this method is called outside a OVS DB transaction | 13:38 |
| ralonsoh | this is not correct | 13:38 |
| sahid | lajoskatona: cool thank you let me see what i can do with fullstack ++ | 13:40 |
| ralonsoh | sean-k-mooney, sorry, I had an old version. The method is now called inside a DB tnx | 13:41 |
| lajoskatona | sahid: the functional framework not really my area, but there you can have tests for modules like for dhcp, and my idea was that as you changed how dhcp works for example worth to check it if you can add there higher level tests | 13:44 |
| lajoskatona | sahid: please check it, my fear is that without some higher level test it is easier to break this feature later | 13:44 |
| lajoskatona | sahid: and the ultimate one is to add tempest test for it, in that I can help with review also (in other things of course also) | 13:45 |
| sean-k-mooney | ralonsoh: yep that was fixed | 13:59 |
| sean-k-mooney | sahid: when working on multi segment things keep in mind that nova does not supprot networks that span physnets | 14:00 |
| sahid | lajoskatona: i understand your point no worries, basically i tried to work incrementally with several patches on that serie so for each step we can see tests running and have eyes focus on | 14:00 |
| sahid | but i will do my best to add tests | 14:01 |
| sean-k-mooney | i.e. nova does not supprot https://github.com/openstack/neutron-lib/blob/master/neutron_lib/api/definitions/multiprovidernet.py | 14:01 |
| *** Guest202 is now known as dasm | 14:03 | |
| sean-k-mooney | so we support multi segment networks with at most 1 physnet per network. | 14:03 |
| sahid | sean-k-mooney: hum we are currently working on multiple segments per network per physnet, there are no desire having more than one physnet for a given provider network at that point, does that make sense? | 14:06 |
| sahid | the only issue that i have noticed so far in nova is reported here https://bugs.launchpad.net/nova/+bug/1979959 | 14:07 |
| sean-k-mooney | sahid: people have tried to use multiple phsnets per network in the past | 14:08 |
| sahid | basically during boot time is you set a fixed ip, there is nothing that select the right segment which handle the subnet of the requested ip | 14:08 |
| sean-k-mooney | multiple segments per network all on the same phsynet is fine | 14:08 |
| sahid | sean-k-mooney: ok cool perfect :-) | 14:08 |
| sean-k-mooney | sahid: ya so the orginal segment extention did not use phsnets at all | 14:09 |
| sean-k-mooney | when we were drafting the spc | 14:09 |
| sean-k-mooney | *spec | 14:09 |
| sean-k-mooney | but the ablitly to assocate networks with hosts was removed | 14:09 |
| sean-k-mooney | *segments with hosts | 14:09 |
| sean-k-mooney | the only way to do that today is via a physnet | 14:10 |
| sahid | yes using physnet and aggregate i guess | 14:10 |
| sean-k-mooney | ya so that does not really work | 14:14 |
| sean-k-mooney | it does | 14:14 |
| sean-k-mooney | but the problem is that if you say have 2 cells | 14:15 |
| sean-k-mooney | one at a central side and one at an edge site | 14:15 |
| sean-k-mooney | the usecase that does not work propelry today | 14:15 |
| sean-k-mooney | is having one segment for the central site and another of the edge | 14:16 |
| sean-k-mooney | today you have to use the same physnet for both if you want sriov/hardwoar offload to work | 14:16 |
| sean-k-mooney | people have tried to use differnt phsynets for the edge and cental sites in the past | 14:17 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: DNM Test FT timeout with oslo master https://review.opendev.org/c/openstack/neutron/+/863315 | 14:19 |
| sahid | sean-k-mooney: i see, thank you for the explanation | 14:20 |
| sean-k-mooney | the original proposal at the first ptg where network segments was designed had a api in neutron to allow you to map segments to hosts. im not sure why we did not take that approch and instead took a config driven approch | 14:22 |
| sean-k-mooney | sahid: im in a meetign for the next few hours but how are you planning to allow mapping segments to hosts | 14:22 |
| sahid | sean-k-mooney: (no worries tkae you time) The original model was already handling multiple segments per network, so basically there was nothing to touch in the server side, only the ovs agent (the one that we are focused on) and dhcp agent have to be updated | 14:26 |
| sahid | The full impl is here https://review.opendev.org/q/topic:segments+-status:abandoned | 14:26 |
| sean-k-mooney | ack ya you could have multiple segments before yes | 14:26 |
| sahid | yep | 14:27 |
| sean-k-mooney | say one vxlan segment and one vlan segment | 14:27 |
| sean-k-mooney | that worked before | 14:27 |
| sean-k-mooney | 2 vlan segments i think did not but cant rememebr | 14:27 |
| sean-k-mooney | or maybe it was 2 vxlan segments | 14:27 |
| sahid | i think this was working but not for a network attached to a physnet | 14:27 |
| sean-k-mooney | the docs were not amazing | 14:27 |
| sean-k-mooney | ya so im pretty sure multipel tunneled segments works fine with ml2/ovs today | 14:28 |
| sean-k-mooney | or ml2/ovn | 14:28 |
| sean-k-mooney | whewn you start to use non tunngled segmetns it didnt work well | 14:28 |
| sean-k-mooney | *multiple non tunneled segments | 14:28 |
| sahid | honeslty i have not tried | 14:30 |
| sean-k-mooney | its been a few years since i did. i tested this when it first merged but i have not touched it in like 4 years or more | 14:30 |
| sahid | back to ovs and the support of multiseg, i was suprised to see the number of people which told me that they use segments, and they have desire for a multiseg support, I tought that no one was really using segments :D | 14:33 |
| sahid | s/tought/thought | 14:34 |
| sean-k-mooney | we recently added schduler support | 14:34 |
| sean-k-mooney | https://specs.openstack.org/openstack/nova-specs/specs/wallaby/implemented/routed-networks-scheduling.html | 14:34 |
| sahid | yes I think it's Sylvain who aorked on it right? | 14:35 |
| sahid | yes it's him, i checked the code and noticed some changes | 14:35 |
| sean-k-mooney | yes | 14:38 |
| opendevreview | Merged openstack/neutron stable/victoria: Revert "[OVN] Set NB/SB "connection" inactivity probe" https://review.opendev.org/c/openstack/neutron/+/863073 | 15:27 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: DNM Test FT timeout with oslo master https://review.opendev.org/c/openstack/neutron/+/863315 | 15:53 |
| *** jlibosva is now known as Guest257 | 17:08 | |
| *** labedz is now known as labedz_ | 17:43 | |
| *** elvira1 is now known as elvira | 18:34 | |
| *** dasm is now known as dasm|off | 21:52 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!