opendevreview | Elvira García Ruiz proposed openstack/neutron master: Make OVN driver validate Geneve max_header_size https://review.opendev.org/c/openstack/neutron/+/714716 | 00:03 |
---|---|---|
opendevreview | Merged openstack/python-neutronclient master: Use yaml.safe_load instead of yaml.load https://review.opendev.org/c/openstack/python-neutronclient/+/801224 | 02:27 |
opendevreview | Manu B proposed openstack/neutron-dynamic-routing master: Use payload for PORT AFTER_UPDATE event https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/806543 | 03:29 |
opendevreview | Manu B proposed openstack/neutron-dynamic-routing master: Use payload for PORT and FLOATING_IP events https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/806543 | 03:31 |
opendevreview | Merged openstack/neutron master: Do not fail if the agent load is not bumped https://review.opendev.org/c/openstack/neutron/+/804218 | 04:04 |
opendevreview | Manu B proposed openstack/neutron-dynamic-routing master: Support router association to BGP speaker https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/792338 | 05:17 |
opendevreview | Manu B proposed openstack/neutron-dynamic-routing master: Support peer association to BGP speaker https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/800882 | 05:47 |
opendevreview | Elvira García Ruiz proposed openstack/neutron master: Make OVN driver validate Geneve max_header_size https://review.opendev.org/c/openstack/neutron/+/714716 | 08:41 |
opendevreview | Kevin Li proposed openstack/neutron master: update subport status when trunk/subport create/delete is triggerred https://review.opendev.org/c/openstack/neutron/+/807076 | 09:02 |
bcafarel | ralonsoh: hi, small easy one for your backlog https://review.opendev.org/c/openstack/networking-sfc/+/782790 (automated patch we forgot about at wallaby creation) | 09:03 |
ralonsoh | bcafarel, for sure | 09:04 |
bcafarel | thanks :) | 09:05 |
opendevreview | Kevin Li proposed openstack/neutron master: update subport status when trunk/subport create/delete is triggerred https://review.opendev.org/c/openstack/neutron/+/807076 | 09:07 |
slaweq | ralonsoh: hi, can You maybe +W https://review.opendev.org/c/openstack/neutron/+/806746 ? | 09:20 |
slaweq | thx in advance | 09:20 |
ralonsoh | slaweq, let me check | 09:20 |
slaweq | thx | 09:20 |
ralonsoh | done | 09:20 |
opendevreview | Manu B proposed openstack/neutron-dynamic-routing master: Use payload for PORT and FLOATING_IP events https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/806543 | 09:50 |
opendevreview | Manu B proposed openstack/neutron-dynamic-routing master: use payloads for ROUTER_INTERFACE events https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/639852 | 09:50 |
opendevreview | Merged openstack/neutron master: Use "objects_exist" in PortForwardingPlugin https://review.opendev.org/c/openstack/neutron/+/806962 | 10:08 |
opendevreview | Merged openstack/neutron master: [Doc] Fix link to the Mellanox documentation in SR-IOV doc https://review.opendev.org/c/openstack/neutron/+/806860 | 10:08 |
opendevreview | Merged openstack/networking-sfc master: Update master for stable/wallaby https://review.opendev.org/c/openstack/networking-sfc/+/782790 | 10:17 |
opendevreview | Rodolfo Alonso proposed openstack/neutron master: Add port status to information dictionary in agent RPC https://review.opendev.org/c/openstack/neutron/+/806889 | 10:35 |
opendevreview | Merged openstack/networking-sfc master: Add Python3 xena unit tests https://review.opendev.org/c/openstack/networking-sfc/+/782791 | 10:36 |
opendevreview | Mamatisa Nurmatov proposed openstack/neutron master: Refactor notify to publish https://review.opendev.org/c/openstack/neutron/+/807114 | 10:51 |
opendevreview | Oleg Bondarev proposed openstack/neutron master: Local IP RPC server-agent interface https://review.opendev.org/c/openstack/neutron/+/807116 | 10:58 |
opendevreview | Manu B proposed openstack/neutron-dynamic-routing master: Push router,peer events to BGP l3 agent extension https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/802445 | 11:02 |
opendevreview | Lajos Katona proposed openstack/neutron-tempest-plugin master: WIP: API tests for BFD support https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/800949 | 11:30 |
ralonsoh | lajoskatona, slaweq about the problem with neutron.tests.fullstack.test_l3_agent.TestHAL3Agent.test_router_fip_qos_after_admin_state_down_up | 11:36 |
ralonsoh | I think we have, again, a problem with the DHCP agent | 11:36 |
ralonsoh | the port to be added to the router is never provisioned by any DHCP agent | 11:36 |
slaweq | ralonsoh: so the issue which obondarev was trying to solve some time ago, right? | 11:38 |
ralonsoh | yes | 11:38 |
slaweq | by adding more dhcpagents in the test | 11:38 |
slaweq | damm :/ | 11:38 |
slaweq | thx for triaging it ralonsoh | 11:38 |
ralonsoh | is there a bug? | 11:38 |
ralonsoh | a LP bug | 11:38 |
slaweq | 1sec | 11:38 |
slaweq | https://bugs.launchpad.net/neutron/+bug/1942190 | 11:39 |
ralonsoh | slaweq, thanks! | 11:39 |
slaweq | that's the one which I opened this week | 11:39 |
slaweq | if You want the old one, I can look for it | 11:39 |
ralonsoh | I'll do it | 11:39 |
slaweq | thx | 11:40 |
lajoskatona | This one I think: https://bugs.launchpad.net/neutron/+bug/1930401 | 11:41 |
ralonsoh | yes, this one | 11:41 |
lajoskatona | for this we changed oslo.privsep to have timeout, but seems that is not enough | 11:41 |
lajoskatona | https://review.opendev.org/c/openstack/neutron/+/794994 | 11:42 |
ralonsoh | lajoskatona, e.g.: https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_82c/803462/2/check/neutron-fullstack-with-uwsgi/82c9760/controller/logs/dsvm-fullstack-logs/TestHAL3Agent.test_router_fip_qos_after_admin_state_down_up/neutron-dhcp-agent--2021-08-25--14-57-50-090655_log.txt | 11:42 |
lajoskatona | ralonsoh, slaweq: personal warning: probably I will disappear again from IRC, I am on a not reliable network.... | 11:43 |
ralonsoh | np | 11:44 |
slaweq | lajoskatona: sure, thx for the heads up | 11:44 |
obondarev | ralonsoh: looks like in last case 1 of 2 dhcp agents hanged and the second one does not host the net. So port was not handled | 11:49 |
ralonsoh | obondarev, exactly | 11:50 |
obondarev | wondering if updating config for number of agents hosting net could help | 11:50 |
obondarev | but again only a workaround | 11:50 |
ralonsoh | yes, this is a workaround until we can fix the privsep problems | 11:52 |
obondarev | lajoskatona: the patch dealing with privsep was not landed, right lajoskatona? | 11:52 |
ralonsoh | nope | 11:52 |
lajoskatona | no | 11:52 |
ralonsoh | in any case, I'll try your patch for HA too | 11:52 |
lajoskatona | ralonsoh: ok | 11:53 |
opendevreview | Merged openstack/neutron master: Remove dhcp_extra_opt value after first newline character https://review.opendev.org/c/openstack/neutron/+/806746 | 12:16 |
*** frenzy_friday is now known as anbanerj|ruck | 12:46 | |
opendevreview | Rodolfo Alonso proposed openstack/neutron master: Use 2 more DHCP agents in TestHAL3Agent https://review.opendev.org/c/openstack/neutron/+/807134 | 13:06 |
mloza | hello, i'm playing stateless security group and i understand you cannot mix stateless and stateful on the same port but what if I have two ports, one is stateful and other one is stateless, will this break things or should work? | 14:17 |
opendevreview | Rodolfo Alonso proposed openstack/neutron master: Use 2 DHCP agents per network in TestHAL3Agent https://review.opendev.org/c/openstack/neutron/+/807134 | 14:27 |
opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN][Placement] Add a SB Chassis event to track changes in BW config https://review.opendev.org/c/openstack/neutron/+/801712 | 14:31 |
opendevreview | Merged openstack/neutron master: Bump neutron-lib to 2.13.0 https://review.opendev.org/c/openstack/neutron/+/803462 | 14:39 |
opendevreview | Mamatisa Nurmatov proposed openstack/neutron master: Rename notify to publish in unit tests https://review.opendev.org/c/openstack/neutron/+/807114 | 14:41 |
opendevreview | Slawek Kaplonski proposed openstack/neutron master: Add pagging and sorting support for "network_ip_availability" https://review.opendev.org/c/openstack/neutron/+/806404 | 14:49 |
opendevreview | Slawek Kaplonski proposed openstack/neutron master: Add pagging and sorting support for "network_ip_availability" https://review.opendev.org/c/openstack/neutron/+/806404 | 14:51 |
opendevreview | Merged openstack/neutron stable/wallaby: Remove dhcp_extra_opt value after first newline character https://review.opendev.org/c/openstack/neutron/+/806748 | 14:53 |
elvira | ralonsoh: lucasagomes bcafarel and any other reviewers that might be interested. I've got both https://review.opendev.org/c/openstack/neutron/+/800278 and https://review.opendev.org/c/openstack/neutron/+/714716 ready to be reviewed, just if you have time | 15:01 |
ralonsoh | elvira, let me check | 15:01 |
elvira | thanks a lot! | 15:01 |
opendevreview | Merged openstack/neutron-specs master: Create intermediate OVS bridge to improve live-migration in OVN https://review.opendev.org/c/openstack/neutron-specs/+/799198 | 15:03 |
opendevreview | Slawek Kaplonski proposed openstack/neutron master: Use ovs constants from neutron-lib https://review.opendev.org/c/openstack/neutron/+/797120 | 15:05 |
lajoskatona | slaweq, ralonsoh, obondarev: Could You check the n-d-r patch series (2 patches only) starting with this one: https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/806543 | 15:37 |
ralonsoh | sure | 15:37 |
obondarev | yep | 15:38 |
lajoskatona | thanks | 15:39 |
obondarev | ralonsoh, please check https://review.opendev.org/c/openstack/neutron-lib/+/804613/ when have some time - you've seen this already, I just rebased and added 1 more constant | 15:47 |
ralonsoh | sure | 15:47 |
obondarev | thanks! | 15:47 |
opendevreview | Terry Wilson proposed openstack/neutron master: Support SB OVSDB connections to non-leader servers https://review.opendev.org/c/openstack/neutron/+/803268 | 15:55 |
bcafarel | elvira: sure, looking | 16:05 |
opendevreview | Rodolfo Alonso proposed openstack/neutron master: Use 2 DHCP agents per network in TestHAL3Agent https://review.opendev.org/c/openstack/neutron/+/807134 | 16:05 |
opendevreview | Terry Wilson proposed openstack/neutron master: Use lookup() instead of iterating over table https://review.opendev.org/c/openstack/neutron/+/802607 | 16:07 |
elvira | bcafarel: thanks!! | 16:08 |
opendevreview | Merged openstack/neutron stable/wallaby: Do not fail if the agent load is not bumped https://review.opendev.org/c/openstack/neutron/+/806568 | 16:15 |
opendevreview | Merged openstack/neutron stable/victoria: Do not fail if the agent load is not bumped https://review.opendev.org/c/openstack/neutron/+/806569 | 16:16 |
opendevreview | Merged openstack/neutron stable/ussuri: Do not fail if the agent load is not bumped https://review.opendev.org/c/openstack/neutron/+/806571 | 16:27 |
opendevreview | Merged openstack/neutron-dynamic-routing master: Use payload for PORT and FLOATING_IP events https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/806543 | 16:34 |
opendevreview | Merged openstack/neutron-dynamic-routing master: use payloads for ROUTER_INTERFACE events https://review.opendev.org/c/openstack/neutron-dynamic-routing/+/639852 | 16:34 |
opendevreview | Merged openstack/neutron stable/wallaby: ovn: Consider all router ports in is_lsp_router_port() https://review.opendev.org/c/openstack/neutron/+/806866 | 16:35 |
opendevreview | Merged openstack/neutron stable/ussuri: ovn: Consider all router ports in is_lsp_router_port() https://review.opendev.org/c/openstack/neutron/+/806869 | 16:35 |
opendevreview | Merged openstack/neutron stable/victoria: ovn: Consider all router ports in is_lsp_router_port() https://review.opendev.org/c/openstack/neutron/+/806868 | 16:35 |
spatel | ralonsoh is it possible my floating use SRIOV? | 16:46 |
spatel | also VxLAN tunnel traffic run on SRIOV also? | 16:47 |
ralonsoh | spatel, sorry what? | 16:47 |
spatel | I want high performance networking throughput in my openstack network layer | 16:47 |
spatel | Example tenant traffic run on SRIOV or DPDK using vxlan or genev tunnel. (something like that) | 16:48 |
spatel | is it possible ? i am new in that design so asking question myself | 16:48 |
ralonsoh | SRIOV only supports flat and VLAN | 16:49 |
ralonsoh | https://github.com/openstack/neutron/blob/master/neutron/plugins/ml2/drivers/mech_sriov/mech_driver/mech_driver.py#L89 | 16:49 |
ralonsoh | not tunneled networks | 16:49 |
spatel | ah! make sense | 16:50 |
ralonsoh | and about having a FIP with a SRIOV port, doesn't make sense | 16:50 |
spatel | what about DPDK ? | 16:50 |
ralonsoh | why would you do DNAT on a port that should be connected directly to a external switch? | 16:50 |
spatel | does that support tunnel and FIP? | 16:50 |
ralonsoh | OVS+DPDK? | 16:51 |
ralonsoh | OVN+DPDK? | 16:51 |
spatel | Yes OVS+OVN + DPDK | 16:51 |
ralonsoh | with OVN, if you use DVR (this is the default config), I makes sense to have FIP if you need | 16:52 |
ralonsoh | that will be DNATed inside the same OVS | 16:52 |
spatel | i am just exploring all option for my nest new cloud, requirement is to have high network throughput and low latency | 16:52 |
ralonsoh | with OVS you'll need to send it to iptables | 16:52 |
ralonsoh | you loose the advantage | 16:53 |
spatel | >> with OVS you'll need to send it to iptables ?? | 16:53 |
ralonsoh | this is how fixed IP is DNATed to FIP | 16:54 |
spatel | not for OVN right? | 16:59 |
ralonsoh | OVN implements everything with OpenFlow rules. With DVR, this DNAT will be done inside the same OVS instance, in the same datapath | 17:00 |
spatel | In short SRIOV not supported for tunnel but DPDK does support right? | 17:00 |
ralonsoh | not DPDK, but OVS or OVN | 17:01 |
ralonsoh | DPDK is just a particular implementation of OVS | 17:01 |
spatel | hmm! then how does people run high performance networking for inter VM communication? | 17:01 |
ralonsoh | ? | 17:02 |
ralonsoh | what does it means? | 17:02 |
spatel | If i create VPC and spin up two vm1 and vm2 on two different compute node then they will use VxLAN tunnel to communicate with each other right? | 17:03 |
spatel | if vm1 sending lots of traffic to vm2 then there will be poor performance if they use tap interface because all packet processing happened in Linux kernel.. | 17:03 |
spatel | am i wrong here? | 17:03 |
ralonsoh | depending on how you want to send the traffic | 17:04 |
ralonsoh | if you are using tenant networks | 17:04 |
ralonsoh | and those networks use VLAN | 17:04 |
spatel | I am talking about tenant traffic | 17:04 |
ralonsoh | you can use (1) SRIOV, the traffic will go from a SRIOV port to another SRIOV port | 17:05 |
ralonsoh | (2) OVN+DPDK or OVS+DPDK if you don't have SRIOV | 17:05 |
spatel | But again question is does DPDK solution work with tenant traffic? | 17:05 |
ralonsoh | this is not a yes or not answer | 17:06 |
ralonsoh | for OVS with VLAN traffic, the physical bridges will be connected to NICs | 17:07 |
ralonsoh | those NICs will be in the same L2 domain | 17:07 |
ralonsoh | those NICs will use DPDK ports --> that will be valid for high speed | 17:08 |
ralonsoh | with OVS and tunneled traffic, there is a method to drive the tunneled traffic trhough the physical bridges | 17:08 |
spatel | That is what i have in my current datacenter, all my VM directly connected to L2 domain of my physical switch fabric and i don't have any tenant traffic and life is good | 17:08 |
ralonsoh | that provides the same high speed | 17:08 |
spatel | >> with OVS and tunneled traffic, there is a method to drive the tunneled traffic trhough the physical bridges - This is i want to know and deploy in new datacenter | 17:09 |
ralonsoh | I don't know if this is officially supported, but there is a project that provides this | 17:09 |
ralonsoh | https://opendev.org/x/networking-ovs-dpdk | 17:10 |
spatel | hmm if i attach my vxlan nic directly to sriov or dpdk then it should work right? | 17:10 |
spatel | (vxlan bridge i meant) | 17:10 |
ralonsoh | not at all | 17:10 |
ralonsoh | again, with SRIOV you don't have tunneled traffic | 17:11 |
ralonsoh | and with dpdk you need to redirect the tunneled traffic trough a physical bridge | 17:11 |
spatel | I got your point SRIOV is like physical interface | 17:11 |
spatel | Let me play with DPDK to see if i can utilize it with tenant traffic. (may need to run some experiment) | 17:12 |
spatel | I have build 5 node OVN lab, and noticed lots of flows got installed by default on each compute nodes, I am having hard time to understand some of them. | 17:14 |
spatel | Is there any doc about what are these flows and what is the use of them.. ? | 17:14 |
spatel | does more flows create any performance issue with OVS? because packet need to travel each flow table | 17:15 |
opendevreview | Merged openstack/neutron master: Use payload callback for RPC resource_cache https://review.opendev.org/c/openstack/neutron/+/806607 | 17:18 |
opendevreview | Merged openstack/neutron master: Remove IDL classes implemented in ovsdbapp https://review.opendev.org/c/openstack/neutron/+/806437 | 17:18 |
spatel | ralonsoh any idea about that? | 17:22 |
ralonsoh | sorry, I'm with a customer right now | 17:22 |
spatel | no worry, take your time | 17:23 |
opendevreview | Merged openstack/neutron master: Enable QoS min bw tempest in neutron-ovs-tempest-multinode-full https://review.opendev.org/c/openstack/neutron/+/806170 | 17:29 |
opendevreview | Merged openstack/neutron-tempest-plugin master: Use default "ds512M" advance image in older stable branches (II) https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/806957 | 17:29 |
opendevreview | Sebastian Lohff proposed openstack/neutron master: Fix dynamic segment allocation race condition https://review.opendev.org/c/openstack/neutron/+/788714 | 17:37 |
opendevreview | Merged openstack/neutron master: Improve content of FloatingIP AFTER callbacks https://review.opendev.org/c/openstack/neutron/+/798009 | 17:54 |
opendevreview | Merged openstack/neutron stable/victoria: Remove dhcp_extra_opt value after first newline character https://review.opendev.org/c/openstack/neutron/+/806749 | 17:54 |
opendevreview | Merged openstack/ovsdbapp stable/train: Actually close the connection in Connection.stop() https://review.opendev.org/c/openstack/ovsdbapp/+/796075 | 18:11 |
opendevreview | Merged openstack/neutron master: [OVN] Make external ports aware of network AZs https://review.opendev.org/c/openstack/neutron/+/762550 | 19:03 |
opendevreview | Arkady Shtempler proposed openstack/neutron-tempest-plugin master: Refactoring - Use existing Tempest APIs in "QoS bandwidth limit rule" tests https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/806977 | 19:34 |
opendevreview | Merged openstack/neutron-lib master: Add Local IP constants and exceptions https://review.opendev.org/c/openstack/neutron-lib/+/804613 | 19:57 |
opendevreview | Merged openstack/neutron master: Do not fail when releasing a quota reservation https://review.opendev.org/c/openstack/neutron/+/805031 | 20:44 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!