*** yamamoto has quit IRC | 00:09 | |
*** armax_ has joined #openstack-neutron-ovn | 00:17 | |
*** armax has quit IRC | 00:21 | |
*** armax_ is now known as armax | 00:21 | |
*** salv-orl_ has joined #openstack-neutron-ovn | 00:53 | |
*** yamamoto has joined #openstack-neutron-ovn | 00:55 | |
*** salv-orlando has quit IRC | 00:56 | |
*** salv-orl_ has quit IRC | 00:57 | |
*** armax has quit IRC | 02:13 | |
*** azbiswas has joined #openstack-neutron-ovn | 03:10 | |
*** armax has joined #openstack-neutron-ovn | 03:23 | |
*** fzdarsky_ has joined #openstack-neutron-ovn | 06:56 | |
*** armax has quit IRC | 07:12 | |
*** azbiswas has quit IRC | 07:19 | |
*** azbiswas has joined #openstack-neutron-ovn | 07:24 | |
*** yamamoto has quit IRC | 09:26 | |
*** yamamoto has joined #openstack-neutron-ovn | 10:26 | |
*** yamamoto has quit IRC | 10:32 | |
*** yamamoto has joined #openstack-neutron-ovn | 11:03 | |
*** yamamoto has quit IRC | 11:03 | |
openstackgerrit | Merged openstack/networking-ovn: Add subtransactions=True on transaction for delete_port/network https://review.openstack.org/223844 | 11:04 |
---|---|---|
*** yamamoto has joined #openstack-neutron-ovn | 12:04 | |
*** yamamoto has quit IRC | 12:09 | |
*** yamamoto has joined #openstack-neutron-ovn | 12:50 | |
*** salv-orlando has joined #openstack-neutron-ovn | 13:31 | |
*** jimchou has joined #openstack-neutron-ovn | 13:46 | |
*** openstackgerrit has quit IRC | 14:31 | |
*** openstackgerrit has joined #openstack-neutron-ovn | 14:31 | |
*** openstackgerrit has quit IRC | 15:01 | |
*** openstackgerrit has joined #openstack-neutron-ovn | 15:01 | |
*** yamamoto has quit IRC | 15:24 | |
*** armax has joined #openstack-neutron-ovn | 15:40 | |
*** yamamoto has joined #openstack-neutron-ovn | 16:25 | |
*** yamamoto has quit IRC | 16:29 | |
openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: test https://review.openstack.org/205187 | 16:36 |
openstackgerrit | Gal Sagie proposed openstack/networking-ovn: Stop logging deadlock tracebacks https://review.openstack.org/224193 | 16:49 |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: Test security groups. https://review.openstack.org/223839 | 16:52 |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: wip: security group support https://review.openstack.org/223817 | 16:52 |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: Temporarily default to ovn-acl branch https://review.openstack.org/223816 | 16:52 |
russellb | gsagie mestery not bad review time on https://review.openstack.org/#/c/224193/1 | 16:55 |
mestery | lol | 16:55 |
mestery | :)\ | 16:55 |
russellb | less than 200 new lines of code to implement security groups ... not bad. | 16:56 |
russellb | net new that is | 16:56 |
russellb | for neutron ... and then only about 250 to add it to OVN | 16:57 |
*** gsagie_ has joined #openstack-neutron-ovn | 17:01 | |
azbiswas | @russellb: did you get a chance to see my 50000 feet ovn-vtep workflow understanding | 17:13 |
*** yamamoto has joined #openstack-neutron-ovn | 17:13 | |
openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: Raise idlutils.RowNotFound over RuntimeError and handle that in plugin https://review.openstack.org/224225 | 17:58 |
openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: Raise idlutils.RowNotFound over RuntimeError and handle in plugin https://review.openstack.org/224225 | 17:58 |
armax | ping actually | 18:08 |
armax | question for the ovn folks | 18:08 |
armax | it seems my devstack got screwed after I tried to stack on ovn | 18:08 |
armax | if I switch back to ml2 my ovs is totally busted | 18:09 |
armax | is that a strategy to let me give up ml2 on ovs agent? | 18:09 |
armax | jokes aside, has anyone encountered that? | 18:10 |
gsagie_ | Yes, i did | 18:10 |
armax | gsagie: sweet | 18:10 |
gsagie_ | armax: i have to re-compile OVS 2.3 for that to work back | 18:11 |
gsagie_ | had | 18:11 |
*** Guest91493 has joined #openstack-neutron-ovn | 18:11 | |
armax | what If uninstall it entirely? | 18:11 |
gsagie_ | I guess that will work too | 18:11 |
armax | apparently it didn’t…I’ll have to find out more what’s going on | 18:12 |
gsagie_ | But you need to make sure to re install OVS 2.3 because OVN plugin currently uninstall it | 18:12 |
gsagie_ | and re-load the kernel modules | 18:12 |
armax | ok | 18:13 |
gsagie_ | If i remember correctly the problem was something with the db schema | 18:13 |
armax | gsagie_: ok, let me give it a go, I actually didn’t spend too much time int oit | 18:13 |
gsagie_ | armax: np, let me know if you need any help because i was stuck on it as well | 18:14 |
armax | cool | 18:14 |
armax | I was also nothing something funny | 18:14 |
armax | which I was going to validate…and file a bug in case my tests were positive | 18:15 |
armax | in a nutshell I cannot delete a network if I have dhcp ports on it | 18:15 |
gsagie_ | But we really do need to change this installation process | 18:15 |
armax | gsagie_: sure | 18:15 |
gsagie_ | armax: yes, i think i know what is your problem | 18:16 |
gsagie_ | but i think it was actually solved so i am suprised | 18:16 |
*** openstackgerrit has quit IRC | 18:16 | |
armax | gsagie_: ok, I might be running on an old version | 18:16 |
gsagie_ | basically because the DHCP port delete is never being sent to the plugin its kept in the NB DB | 18:16 |
armax | gsagie_: of the plugin | 18:16 |
*** openstackgerrit has joined #openstack-neutron-ovn | 18:16 | |
armax | gsagie_: so long as you’re aware I am not going to drill down into it | 18:17 |
gsagie_ | so there might be a problem that the DB is refusing to delete the lswitch | 18:17 |
gsagie_ | armax: you can open a bug and assign it to me and i will verify tommorow | 18:17 |
armax | ok | 18:17 |
gsagie_ | thanks | 18:17 |
mestery | armax: Can't you just fire off a new VM in the HP Cloud and use one for OVN, one for ML2, etc? | 18:21 |
* mestery ducks | 18:21 | |
armax | mestery: wow, I hadn’t thought of that | 18:21 |
mestery | armax: hi | 18:21 |
armax | :) | 18:21 |
mestery | lol | 18:21 |
armax | hi | 18:21 |
mestery | Kidding aside, it is really slick to do that. | 18:21 |
armax | to do what? | 18:21 |
openstackgerrit | Merged openstack/networking-ovn: Stop logging deadlock tracebacks https://review.openstack.org/224193 | 18:22 |
armax | using HP Cloud? | 18:22 |
mestery | yes | 18:22 |
armax | but I have my pet | 18:22 |
Guest91493 | armax: yea i'm not sure if you heard of this new thing called cloud ? | 18:22 |
armax | Guest91493: identify yourself | 18:22 |
*** Guest91493 has quit IRC | 18:22 | |
*** Guest91493 has joined #openstack-neutron-ovn | 18:22 | |
Guest91493 | hi | 18:22 |
armax | cloud? | 18:22 |
*** Guest91493 is now known as arosen | 18:22 | |
arosen | boom | 18:22 |
armax | I never heard of it before | 18:22 |
armax | speaks the guy who doesn’t use vpn | 18:22 |
armax | meh | 18:23 |
arosen | icbb | 18:23 |
mestery | MOAR CLOUD! | 18:23 |
armax | and installs on bare metal | 18:23 |
arosen | speaking of which i need to reserve squash | 18:23 |
mestery | VPN? | 18:23 |
mestery | For what? | 18:23 |
armax | do do that | 18:23 |
mestery | cloud? | 18:23 |
mestery | I don't need cloud for that | 18:23 |
mestery | I use floating IPs | 18:23 |
mestery | lol | 18:23 |
armax | FIPs? | 18:24 |
mestery | FIPs | 18:24 |
armax | nah, you gotta connect to the external network | 18:24 |
mestery | moar fips | 18:24 |
armax | external networks are the new black | 18:24 |
arosen | One of my roommates is in sales and sells an HR cloud platform and he says to people, "One of the great about moving to cloud is there is no more code. Everything is in the cloud" | 18:25 |
mestery | https://youtu.be/nTx5xmwHDOw | 18:25 |
arosen | I could only listen to that for about 10 seconds mestery | 18:26 |
mestery | arosen: You're weak :) | 18:26 |
arosen | maybe. | 18:26 |
mestery | lol | 18:26 |
mestery | arosen: A more hi fidelty version: https://open.spotify.com/track/4vTZNpR1iI4BsUDbz4l6dy | 18:27 |
armax | I stoped at second #3 | 18:27 |
armax | is that bad? | 18:27 |
mestery | lol | 18:27 |
armax | my catholic ears started to bleed | 18:27 |
mestery | rofl | 18:27 |
armax | :) | 18:27 |
mestery | They play that at mass every week here in Minnesota | 18:27 |
mestery | :) | 18:27 |
armax | I am going to hell now aren’t I? | 18:28 |
mestery | Well | 18:28 |
mestery | hi | 18:28 |
mestery | :D | 18:28 |
armax | … | 18:29 |
arosen | btw have you guys seen: https://aws.amazon.com/lambda/ | 18:29 |
mestery | OMG | 18:29 |
mestery | arosen: That's the GREATEST THING EVER! | 18:29 |
mestery | Seriously | 18:29 |
arosen | Maybe i'm just super late to the game... I just found out about that yesterday. | 18:29 |
armax | yes you are | 18:29 |
arosen | :) | 18:29 |
mestery | arosen: Similar but not for AWS: http://stackhut.com/#/ | 18:29 |
mestery | arosen: That stuff is the future | 18:29 |
arosen | Yea i saw they released it in 2014 nov | 18:29 |
armax | arosen: mestery knew about that way earlier than you | 18:29 |
mestery | It's AWESOME | 18:29 |
mestery | rofl | 18:29 |
armax | like two weeks ago | 18:29 |
mestery | Well | 18:29 |
mestery | carl_baldwin andI saw a demo at Linuxcon in Seattle | 18:30 |
mestery | We were blown away | 18:30 |
gsagie_ | arosen: lambda for OpenStack: https://review.openstack.org/#/c/164537/ | 18:30 |
arosen | mestery: spends more time on twitter than me so I figured he would :) | 18:30 |
mestery | lol | 18:30 |
mestery | Dude | 18:30 |
mestery | I'm social | 18:30 |
arosen | I found about about it from CloudOpinion | 18:30 |
arosen | that guy is hilarious | 18:30 |
mestery | I love cloudopinion | 18:31 |
arosen | yea he's great | 18:31 |
mestery | Maybe cloudopinion can get armax to tweet more | 18:31 |
armax | me? | 18:31 |
armax | I ain’t got time for that | 18:32 |
mestery | lol | 18:33 |
mestery | Well | 18:33 |
mestery | Should you become PTL, that'll be 100% right :D | 18:33 |
armax | you talking to me? | 18:35 |
mestery | hi | 18:37 |
openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: test https://review.openstack.org/205187 | 18:49 |
openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: Raise idlutils.RowNotFound over RuntimeError and handle in plugin https://review.openstack.org/224225 | 19:10 |
*** gsagie_ has quit IRC | 19:11 | |
arosen | armax: any idea about these two errors i've been seeing: | 19:11 |
arosen | http://logs.openstack.org/25/224225/2/check/gate-tempest-dsvm-networking-ovn/8719aa6/logs/screen-q-svc.txt.gz?level=TRACE#_2015-09-16_18_32_19_264 | 19:11 |
arosen | http://logs.openstack.org/25/224225/2/check/gate-tempest-dsvm-networking-ovn/8719aa6/logs/screen-q-svc.txt.gz?level=TRACE#_2015-09-16_18_39_49_820 | 19:11 |
arosen | I've googled around a bit but nothing yet hops out. | 19:12 |
arosen | I asked zzzeek in #openstac-dev we'll see if he has any ideas. | 19:12 |
armax | not sure I have seen it before | 19:13 |
arosen | it wasn't happening until you added your config file changes. | 19:14 |
arosen | Just kidding :P | 19:14 |
arosen | yea i haven't seen either of these errors in the past. It doesn't occur consistently and i haven't been able to reproduce it locally. | 19:15 |
*** shettyg has joined #openstack-neutron-ovn | 19:18 | |
* russellb hasn't tried to do a conversion to/from ML2+OVS yet | 19:26 | |
russellb | but not surprised you run into something with devstack | 19:27 |
russellb | even if you reinstall the OVS package, you'll have 2 copies of OVS installed | 19:27 |
russellb | since OVN devstack installs ovs master from source into /usr/local// | 19:27 |
russellb | might be worth a devstack document update ... | 19:27 |
russellb | though I generally assume any devstack config destroys a system :) | 19:28 |
armax | russellb: yes | 19:49 |
armax | russellb: the devstack install doesn’t rollback things the way they were | 19:50 |
russellb | right | 19:50 |
armax | they uninstall the distro ovs | 19:50 |
armax | and install ovs locally | 19:51 |
armax | so the whole things is a bit confused | 19:51 |
armax | we could make the ovn’s unstack more polite | 19:51 |
russellb | and then says, oh you want to *not* use ovs from git? ¯\_(ツ)_/¯ | 19:51 |
armax | but I leave that exercise to the reader | 19:51 |
russellb | :) | 19:52 |
armax | :) | 19:52 |
russellb | yeah, never thought about uninstalling during unstack | 19:52 |
russellb | that's a good idea | 19:52 |
armax | I changed the prefix to /usr | 19:52 |
armax | because I am rude | 19:52 |
armax | reinstalled ovs | 19:52 |
armax | and things went back to normal | 19:53 |
azbiswas | speaking of devstack and ovn, in my ubuntu/trusty64, the ovn stack failed without installing python-openvswitch first. Has anyone encountered this before? | 19:53 |
russellb | azbiswas: it shouldn't ... that should be getting installed via pip | 19:53 |
russellb | not sure why that happened | 19:54 |
russellb | and it looks like you can run "sudo make uninstall" for ovs | 19:54 |
russellb | heck, i'll even *test* it! | 19:54 |
armax | russellb: true | 19:55 |
armax | but when you do reinstall the distro ova | 19:55 |
armax | ov | 19:55 |
armax | ovs | 19:55 |
armax | the config path is now pointing to /usr/local | 19:55 |
russellb | which config path | 19:55 |
russellb | sounds like you could have been using ovs-vsctl and friends from /usr/local, which would default to /usr/local | 19:56 |
russellb | and that would break | 19:56 |
armax | yes | 19:56 |
armax | the env gets dirty in a way | 19:56 |
* russellb nods | 19:56 | |
armax | I am happy with the latest ovs/ovn capable thingy | 19:56 |
armax | hence I installed directly on /usr | 19:57 |
armax | I mean by changing the prefix | 19:57 |
armax | on configure | 19:57 |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: devstack: Remove ovs during unstack. https://review.openstack.org/224282 | 20:01 |
russellb | armax: ^ :) | 20:01 |
armax | russellb: we should reinstall the ovs that we uninstalled at line 166 | 20:02 |
armax | otherwise the user will be left high and dry | 20:02 |
armax | no? | 20:02 |
russellb | armax: wellllllllllll | 20:02 |
russellb | it could have been devstack that installed it in the first place | 20:02 |
russellb | it usually is | 20:02 |
*** mestery has quit IRC | 20:02 | |
armax | true | 20:02 |
russellb | usually neutron requirements bring it in, then we promptly uninstall it | 20:03 |
russellb | so *shrug* | 20:03 |
armax | there’s still something funky about the config prefix | 20:03 |
russellb | more complete fix would be to prevent it from getting installed in the first place automatically based on neutron package list, and then only re-install if it was already installed pre-devstack | 20:03 |
russellb | or something | 20:03 |
*** mestery has joined #openstack-neutron-ovn | 20:03 | |
russellb | and that's starting to sound like real work | 20:03 |
armax | yeah | 20:04 |
armax | and no-one ain’t got time for that | 20:04 |
russellb | exactly! | 20:04 |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: devstack: Remove ovs during unstack. https://review.openstack.org/224282 | 20:09 |
azbiswas | russellb: Can we discuss the ovn plugin support for vtep | 20:18 |
russellb | yes | 20:18 |
russellb | my inbox is a disaster so i haven't read your mail yet, sorry | 20:18 |
russellb | i can read now though | 20:18 |
azbiswas | I was trying to see if my workflow for OVN VTEP is correct | 20:18 |
russellb | OK, sure | 20:19 |
russellb | i just read what you had, so let's start at the beginning | 20:19 |
azbiswas | OK | 20:20 |
russellb | first, within OVN, there is ovn-controller-vtep running managing a switch that supports the hardware_vtep schema | 20:20 |
russellb | (it could be running on the switch itself, or somewhere else, doesn't really matter) | 20:20 |
azbiswas | sure | 20:20 |
russellb | an admin of the switch will create logical switches in hardware_vtep | 20:21 |
russellb | those logical switches are associated with VLANs on some subset of the physical ports | 20:21 |
azbiswas | This representation will be pushed to the OVN Southbound | 20:22 |
russellb | not quite | 20:22 |
russellb | that's where we diverge a bit | 20:22 |
russellb | soooo, say I want to connect a Neutron network (OVN_Northbound logical switch) to a logical switch on a vtep gateway | 20:23 |
russellb | <--- insert big question mark about how you tell Neutron to do that ----> | 20:23 |
azbiswas | So the Openstack admin must first create a generic Logical Switch | 20:23 |
russellb | what needs to happen in OVN is that Neutron should create a Logical Port in OVN_Northbound with a type: vtep and options: vtep-physical-switch=<name> vtep-logical-switch=<name> | 20:23 |
*** chandra_v has joined #openstack-neutron-ovn | 20:24 | |
russellb | The connection to the vtep gateway is modeled as another logical port on a logical switch in OVN_Northbound | 20:24 |
russellb | (we can cover what happens after that with ovn-northd and ovn-controller-vtep in a minute) | 20:25 |
azbiswas | I am with you there, the missing part for me is how did the logical switch on a vtep gateway appear in the OVN Northbound | 20:25 |
russellb | Neutron creates is | 20:25 |
russellb | it* | 20:26 |
russellb | oh | 20:26 |
russellb | that's not what you asked | 20:26 |
arosen | armax: i think this fixes the first error https://review.openstack.org/224289 | 20:26 |
russellb | there is no logical switch in OVN_Northbound for the vtep gateway | 20:26 |
russellb | it's slightly confusing because "logical switch" is used in both OVN_Northbound and hardware_vtep | 20:26 |
russellb | but they are different things | 20:26 |
* armax looks | 20:26 | |
russellb | there are logical switches in hardware_vtep, switch admin managed ... connecting an OVN network to one of those is just done by creating a logical port with type=vtep | 20:27 |
azbiswas | Is there a logical switch for vtep gateway in OVN_Southbound | 20:27 |
azbiswas | ? | 20:27 |
russellb | nope | 20:27 |
russellb | from OVN's perspective, it's just an endpoint, and looks like any other port | 20:27 |
russellb | so it has a Port_Binding row | 20:28 |
russellb | and it's modeled in the logical flow table | 20:28 |
russellb | I think ovn-architecture might discuss this ... *looks* | 20:30 |
russellb | see page 8 on http://www.russellbryant.net/ovs-docs/ovn-architecture.7.pdf | 20:30 |
russellb | "Life Cycle of a VTEP gateway" | 20:30 |
russellb | that might explain it better than me | 20:30 |
azbiswas | I will look through that and get back to you. | 20:31 |
russellb | ok | 20:31 |
russellb | that goes into more detail through the process i was just trying to describe | 20:31 |
russellb | but hopefully better since it's official docs :) | 20:31 |
azbiswas | This was useful though, I think I get it now. | 20:32 |
russellb | great! | 20:32 |
russellb | the big missing piece is ... how do we create the type=vtep port in OVN_Northbound when using Neutron | 20:32 |
russellb | and i still haven't looked at networking-l2gw close enough to know if it matches the use case | 20:33 |
azbiswas | I looked through the networking-l2gw code, there are primitives such as "create_gateway" | 20:34 |
azbiswas | I was looking at the interfaces exposed by that plugin | 20:34 |
azbiswas | create_l2_gateway etc. | 20:35 |
*** regXboi has joined #openstack-neutron-ovn | 21:03 | |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: Test security groups. https://review.openstack.org/223839 | 21:04 |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: wip: security group support https://review.openstack.org/223817 | 21:04 |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: Temporarily default to ovn-acl branch https://review.openstack.org/223816 | 21:04 |
*** cascardo has quit IRC | 21:28 | |
*** cascardo has joined #openstack-neutron-ovn | 21:29 | |
*** regXboi has quit IRC | 21:50 | |
openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: Add if_exists arg to all update/delete commands in ovsdb/commands.py https://review.openstack.org/224225 | 21:59 |
*** jimchou has quit IRC | 22:33 | |
openstackgerrit | Aaron Rosen proposed openstack/networking-ovn: Add if_exists arg to all update/delete commands in ovsdb/commands.py https://review.openstack.org/224225 | 22:59 |
*** yamamoto has quit IRC | 23:39 | |
*** salv-orlando has quit IRC | 23:48 | |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: Test security groups. https://review.openstack.org/223839 | 23:59 |
openstackgerrit | Russell Bryant proposed openstack/networking-ovn: wip: security group support https://review.openstack.org/223817 | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!