Monday, 2016-02-08

flaviofto all the Asian (Chinese calendar) folks in this channel: happy new year!!!00:02
*** salv-orl_ has quit IRC00:19
*** allan_h has quit IRC00:22
*** allan_h has joined #openstack-neutron-ovn00:28
*** salv-orlando has joined #openstack-neutron-ovn01:19
*** brad_behle has quit IRC01:25
*** chandrav has joined #openstack-neutron-ovn01:26
*** yamamoto_ has joined #openstack-neutron-ovn01:48
*** salv-orlando has quit IRC01:50
*** chandrav has quit IRC01:52
*** yamamoto_ has quit IRC02:07
*** yamamoto_ has joined #openstack-neutron-ovn02:08
*** yamamoto_ has quit IRC02:10
*** yamamoto_ has joined #openstack-neutron-ovn02:10
*** yamamoto_ has quit IRC02:10
*** yamamoto_ has joined #openstack-neutron-ovn02:58
*** salv-orlando has joined #openstack-neutron-ovn03:10
*** salv-orlando has quit IRC03:13
*** jckasper has joined #openstack-neutron-ovn03:25
*** jckasper has quit IRC04:05
*** jckasper has joined #openstack-neutron-ovn04:06
*** jckasper has quit IRC04:07
*** jckasper has joined #openstack-neutron-ovn04:07
*** chandrav has joined #openstack-neutron-ovn04:14
*** chandrav has quit IRC04:34
*** salv-orlando has joined #openstack-neutron-ovn04:44
*** salv-orlando has quit IRC04:56
*** chandrav has joined #openstack-neutron-ovn05:05
*** chandrav has quit IRC05:07
*** chandrav has joined #openstack-neutron-ovn05:07
*** chandrav has quit IRC05:12
*** numans has joined #openstack-neutron-ovn06:34
*** allan_h has quit IRC06:39
*** salv-orlando has joined #openstack-neutron-ovn06:48
*** salv-orlando has quit IRC06:55
*** gsagie has joined #openstack-neutron-ovn07:12
*** salv-orlando has joined #openstack-neutron-ovn08:07
*** salv-orlando has quit IRC08:12
*** openstackgerrit has quit IRC09:02
*** openstackgerrit has joined #openstack-neutron-ovn09:02
*** salv-orlando has joined #openstack-neutron-ovn09:31
*** salv-orl_ has joined #openstack-neutron-ovn10:40
*** salv-orlando has quit IRC10:43
*** yamamoto_ has quit IRC11:27
*** numans has quit IRC12:03
*** chandrav has joined #openstack-neutron-ovn12:07
*** salv-orl_ has quit IRC12:18
*** yamamoto has joined #openstack-neutron-ovn12:21
*** roeyc has joined #openstack-neutron-ovn12:27
*** rtheis has joined #openstack-neutron-ovn12:30
*** salv-orlando has joined #openstack-neutron-ovn12:42
*** yamamoto has quit IRC12:42
*** yamamoto has joined #openstack-neutron-ovn12:43
*** yamamoto has quit IRC12:48
*** yamamoto has joined #openstack-neutron-ovn12:58
*** jckasper has quit IRC13:05
*** jckasper has joined #openstack-neutron-ovn13:05
*** gal_ has joined #openstack-neutron-ovn13:06
*** jckasper has quit IRC13:10
*** gsagie has quit IRC13:10
*** gal_ is now known as gsagie13:12
*** yamamoto has quit IRC13:27
*** tongli has joined #openstack-neutron-ovn13:38
*** jckasper has joined #openstack-neutron-ovn13:42
*** yamamoto has joined #openstack-neutron-ovn13:58
*** brad_behle has joined #openstack-neutron-ovn14:01
*** gsagie has quit IRC14:05
Sam-I-Amtoday should be a holiday for denverites14:08
*** gsagie has joined #openstack-neutron-ovn14:14
mesterySam-I-Am: Congrats on that amazing defensive win14:15
mesteryWhat a good game14:15
russellbdefensive win is a good way to put it14:16
russellbwow that defense ..14:17
mesteryI know, right?14:17
Sam-I-Amyeah it was pretty intense14:17
tongliguys, come on. people in carolinas hurt pretty bad!14:17
Sam-I-Ammy theory was if we could stop cam's offense early, we'd have a much better chance14:17
russellbtongli: i live in the carolinas :)14:17
Sam-I-Amcuz the denver offense is terrible14:17
russellbyour defense stepped up in a big way14:18
Sam-I-Ami was stuck in traffic for an hour and a half at 10 last night just trying to get back home... which is downtown.14:18
Sam-I-Amthe police pretty much closed off downtown14:18
russellbyou live in denver?14:18
Sam-I-Amthis city was WILD last night14:18
russellbi bet :)14:18
Sam-I-Amhorns blaring, people screaming, people just... everywhere.14:19
Sam-I-Ami've lived in a small college town during a national championship win, but not a big city like this14:19
tongliI went to bed after the third period.14:19
russellbi thought you lived in texas, but i probably just made that up since you worked for rax...14:19
Sam-I-Amevery time i tried to sleep, more fireworks went off14:19
Sam-I-Amrussellb: rax moved me from denver to texas. survived a year, barely. moved back to denver as a remote worker.14:20
russellbtongli: i probably should have14:20
russellbSam-I-Am: ah, got it14:20
Sam-I-Ami dont seem to function well outside of denver14:20
Sam-I-Amthat was my fourth time moving back here14:20
russellbi can relate to wanting to just stay "home"14:20
russellbi'm back to where i grew up working remote14:20
russellbseems unlikely that i'll ever move out of the area14:21
Sam-I-Ami'm trying not to move again. the other moves were not my choice.14:21
* russellb nods14:21
Sam-I-Amthe economy was ummmm fun between 08 and 11, so i got to bounce around a bit.14:21
russellbmy moves were school and first big boy job14:21
russellbyeah those were rough years..14:22
Sam-I-Amdefinitely some ptsd from those times14:23
Sam-I-Amyou dont forget watching good people with families get laid off14:23
russellbme too14:23
russellbyep, i had that14:23
*** yamamoto has quit IRC14:28
Sam-I-Amhmm @ trying to get devstack to use a local repo and branch for enable_plugin14:32
Sam-I-Amits just not a happy camper14:32
*** regXboi has joined #openstack-neutron-ovn14:33
russellbi believe in you :-p14:37
* Sam-I-Am looks for the magic switch14:38
*** salv-orlando has quit IRC14:41
Sam-I-Ami think that worked14:41
russellbMic22: hi!14:47
Mic22the patch is here:
russellbso you're working on splitting ovn nb and sb dbs?14:48
Mic22I splitted ovn dbs from the master ovs db14:48
Mic22because in HA if you have those tight it's painfull14:48
Mic22when you move OVN you don't usually need to move ovs14:49
Mic22especially if you deploy compute nodes tight to controller nodes14:49
* regXboi reads patch14:49
Mic22so I updated the ctl script to run his ovn db and the nbctl and sbctl to point to the proper db14:49
Mic22it's a small patch I tested on my env and it works but I would like some advices14:50
* russellb reading14:50
mesteryMic22: Nice! regXboi: ^^^^14:50
* regXboi reads patch14:51
regXboiwell hold on... this patch doesn't actually split NB and SB yet14:51
russellbsort of a step 1 of that14:51
russellbMic22: this looks like a good start14:51
regXboiah ok... I was thinking this was further along...14:52
russellbMic22: ovn-northd should probably get a --ovnsb-db arg passed as well14:52
russellbMic22: and it would be nice to extend this a little more to run ovsdb-server twice (once for NB and once for SB)14:53
russellband in fact, make it possible to run only one or the other, or both14:53
regXboirussellb: the ptcp port needs to be an enviro variable as well14:53
russellbi imagine the most common case is running both, but separate processes14:53
russellbso i'd be ok with just that (and not support for *only* running one of them)14:53
regXboiso that you can have the NB listen on one port and the SB on a different port14:53
regXboipretty much, just copy what is in 276364 and 276454 and then we can abandon those14:54
russellbMic22: the "proper" way to post this for feedback is to post the patch to the ovs dev mailing list14:54
russellbCheck out the "git-send-email" tool14:54
*** salv-orlando has joined #openstack-neutron-ovn14:55
russellb$ git send-email -114:55
russellband then i can provide complete feedback there14:55
Mic22I will post this with the separation of dbs14:56
Mic22it's a little step forward to be done14:56
russellbOK great14:56
russellbyou'll also need to add a "Signed-off-by" header to the commit message14:57
Mic22so to recap ptcp ort should be env14:57
russellbthere's a bit more about that in
regXboiyes, the vconsole and vsyslogs need to be enviros as well14:58
regXboiand vfile for that matter14:58
regXboibecause we'll need to be able to set the logging from inside the devstack plugin to work with screen (where we can)14:58
Mic22you mean this part -vconsole:emer -vsyslog:err -vfile:info14:58
regXboiMic22: yes, and the -vconsole:off on L5014:59
Mic22but in start_northd this is: set "$@" -vconsole:emer -vsyslog:err -vfile:info14:59
regXboiyes... I'm talking about both15:00
Mic22so step by step for now I will focus on port and divide dbs15:00
Mic22later on we can put this stuff in proper variables15:00
regXboiMic22: ack - sounds good, we can tune the rest in later revisions15:00
gsagieregXboi: one thing that probably needs to be addressed later when separating the DB's are various consistency problems (in addition to the various problems we have now with Neutron-OVN NB DB)15:10
regXboigsagie: ack15:11
gsagieor atleast i dont know how ovn-northd behaves in all these scenarios15:11
gsagieits using monitor or its constantly polling NB? because with the monitor you might lose some events under load and never sync them15:12
* regXboi mutters, well this is ... annoying15:12
gsagieI have a nice solution for that in Dragonflow, we can iterate on it later on, i think at least the Neutron-OVN NB part should work, i dont remember, does OVSDB support compare and swap?15:14
russellbgsagie: using monitor15:14
russellband it would sync things even if it missed events15:14
Mic22this should have the separation properly done15:16
regXboinit: should we be able to start/stop the NB and SB dbs independently?15:17
Mic22i thought about that but the question is why?15:18
regXboiThat's why I marked it as a nit15:18
Mic22when you have northd running you need both15:18
regXboibut yeah, I'd say submit this to the ML15:18
Mic22probably better15:18
regXboirussellb: ok, can you tell me where struct sbrec_logical_flow is defined?15:19
russellbregXboi: what editor do you use?15:19
regXboirussellb: I bounce around :)15:20
regXboiI'm looking at some files in emacs and some in vim as we speak15:20
russellbwell as a side note, look up "ctags" and integration with your editor of choice15:20
russellbit will help with things like this15:20
russellband it's a trick, because that header is auto generated, not in git15:20
regXboiyeah - I've been figuring that out15:21
regXboiI looked through the *.in files15:21
russellbit's automatically generated from ovn/ovn-sb.ovsschema15:21
regXboiI'm trying to fix this incremental flow patch to be more correct15:21
regXboirussellb: is it acceptable to inject a non-column variable into an autodefined structure?15:23
russellbprobably not15:24
Mic22russellb: practically i have to add signed-off-by and send an email in the dev list?15:24
regXboihmm, then that means I can't carry a "dirty" working variable around without changing the db schema and that strikes me as overkill15:24
russellbMic22: yes15:24
Mic22is there an automatic way or do I have to manually write this in the commit message?15:25
russellbMic22: git commit --amend -s15:26
russellbMic22: and make sure you look at what this means (that you agree with the Developer Certificate of Origin in
russellbif you haven't seen it before15:27
tonglican someone take a look at this patch? it makes vagrant uses the configuration file more, and can configure the provider network any size15:29
tonglior any address block.15:29
lrichard"git format-patch" with --signoff works too15:30
Mic22can you guys see if that's ok15:31
Mic22if it is I will send it15:31
openstackgerritlitong01 proposed openstack/networking-ovn: Vagrant: Generify post-config options for DevStack
Mic22russelb: ping15:36
Mic22russellb: ping15:36
*** chandrav has quit IRC15:40
russellbthanks!  i'll take a look and give it a try today15:41
Mic22Tomorrow I will start documenting HA15:42
Mic22just one question ATM I'm using haproxy to have a port on the loopback interface pointin to dbs15:42
regXboirussellb: my turn to ping15:42
Mic22is there any other way (without a VIP) to pass more arguments for example?15:43
openstackgerritlitong01 proposed openstack/networking-ovn: Vagrant: make provider network creation configurable
russellbMic22: i dont' see your patch.  are you subscribed to the ovs dev mailing list?16:08
russellbyou probably have to subscribe before you can post the patch16:09
*** jckasper has quit IRC16:10
*** jckasper has joined #openstack-neutron-ovn16:14
regXboirussellb: I see it16:14
regXboiand am commenting on it...16:15
russellbgmail marked it as spam16:15
regXboibad gmail16:15
russellbovs build broken, bah16:18
regXboirussellb: what broke?16:19
russellbmight just be local issue16:19
*** armax has joined #openstack-neutron-ovn16:20
russellbpatch posted to fix ..16:21
* Sam-I-Am gets on the ovs dev list16:31
*** numans has joined #openstack-neutron-ovn16:32
Sam-I-Ammoarr email16:35
*** allan_h has joined #openstack-neutron-ovn16:35
*** salv-orl_ has joined #openstack-neutron-ovn16:40
*** salv-orlando has quit IRC16:43
Sam-I-Amthat moment you realize you have to rebase every part of a tiny patch16:45
*** armax has quit IRC16:46
openstackgerritMatthew Kassawara proposed openstack/networking-ovn: Add support for net-mtu extension
regXboiMic22: ping - do you have patches for your HA work?16:55
*** mamulsow has joined #openstack-neutron-ovn17:00
*** salv-orl_ has quit IRC17:05
*** salv-orlando has joined #openstack-neutron-ovn17:05
*** yamamoto has joined #openstack-neutron-ovn17:21
*** yamamoto has quit IRC17:22
Mic22regXboi: I will put tomorrow a documentation in place. It's not really about patching It's more about a slightly different installation17:42
Mic22But ATM I have a running HA proof of concept using ceph rbd for sharing the configuration and I will document step by step this way. Probably I will put the ceph part in a separate document17:43
*** yamamoto has joined #openstack-neutron-ovn17:43
Mic22I was just thinking if make it with virtual IP or without. The advantage of having this without is to be able to use boxes on different L2 segments.17:45
Mic22But maybe it complexify without a really need17:46
regXboiMic22: ack17:46
*** roeyc has quit IRC17:49
Mic22russellb: ovs build broken, bah => something to be done to fix?17:53
*** armax has joined #openstack-neutron-ovn17:56
regXboiMic22: IIRC russellb has pushed a patch already to fix18:04
Mic22where I don't see it18:04
Mic22by the way I posted the new patch that has tovn-northd updated to read the new dbs18:05
regXboiIt hasn't made it to me18:05
regXboithe patch discussion starts at
Mic22you are even in cc in the new patch i changed the topic because it was too long as russel noticed18:06
*** armax has quit IRC18:07
*** armax has joined #openstack-neutron-ovn18:12
regXboiMic22: yep, I can see it18:22
Mic22I will go home now It starts to be late here, have a good day!18:23
*** markmcclain has quit IRC18:24
*** markmcclain has joined #openstack-neutron-ovn18:25
*** yamamoto has quit IRC18:26
*** armax has quit IRC18:31
*** armax has joined #openstack-neutron-ovn18:44
openstackgerritRichard Theis proposed openstack/networking-ovn: Document neutron API extensions supported
*** chandrav has joined #openstack-neutron-ovn18:55
*** salv-orl_ has joined #openstack-neutron-ovn18:58
*** salv-orlando has quit IRC19:01
*** s3wong has joined #openstack-neutron-ovn19:13
*** chandrav has quit IRC19:14
openstackgerritMerged openstack/networking-ovn: Add support for net-mtu extension
*** armax has quit IRC19:26
*** cz_ has joined #openstack-neutron-ovn19:32
*** cz_ is now known as c_z19:33
*** chandrav has joined #openstack-neutron-ovn19:34
*** armax has joined #openstack-neutron-ovn19:34
*** armax has quit IRC19:36
openstackgerritRichard Theis proposed openstack/networking-ovn: Document neutron API extensions supported
*** salv-orl_ has quit IRC19:53
*** tongli has quit IRC19:57
*** numans has quit IRC20:00
openstackgerritMerged openstack/networking-ovn: Document neutron API extensions supported
regXboirussellb: do you understand the difference between <FOO>_FOR_EACH and <FOO>_FOR_EACH_TRACKED ?20:06
russellbwe don't use the _TRACKED functionality yet20:07
russellbit lets you iterate over rows that have actually changed20:07
regXboiI'm trying to rewrite the lflow processing in ovn-controller to use _TRACKED and I'm getting an odd result20:07
russellbdid you enable change tracking on the tables/columns of interest?20:08
regXboiI believe I did, yes20:08
regXboiI just did an _add_all for the purposes of this test20:08
regXboiand that maybe where I'm getting nailed20:08
regXboibecause it looks like I'm processing one more row with _TRACKED than without and that row appears to have garbage in it20:09
*** chandrav has quit IRC20:12
*** armax has joined #openstack-neutron-ovn20:13
regXboiyep, at initialization with _FOR_EACH, I'm looking at 12 flows and with _FOR_EACH_TRACKED (and add_all) I'm looking at 13 flows20:17
regXboiso I've goofed something :)20:17
*** armax has quit IRC20:17
*** armax has joined #openstack-neutron-ovn20:18
*** salv-orlando has joined #openstack-neutron-ovn20:27
*** armax has quit IRC20:30
regXboirussellb: ok, next question that I'm banging my head against... how do I dig a specific column out of an ovsdb_idl structure?  or do I have to reparse the ovsdb schema and pull out the table and column via strings?20:32
regXboiand if you'd rather I move to #openvswitch for these questions, I certainly can :)20:33
*** salv-orlando has quit IRC20:36
russellbsure let's use #openvswitch20:39
* Sam-I-Am wonders where his ovs channel went20:49
*** c_z has quit IRC20:59
*** armax has joined #openstack-neutron-ovn20:59
*** salv-orlando has joined #openstack-neutron-ovn21:01
*** chandrav has joined #openstack-neutron-ovn21:06
*** gangil has joined #openstack-neutron-ovn21:41
*** gangil has joined #openstack-neutron-ovn21:41
*** regXboi has quit IRC22:01
*** rtheis has quit IRC22:07
*** jckasper has quit IRC22:53
*** jckasper has joined #openstack-neutron-ovn22:54
*** jckasper has quit IRC22:58
*** allan_h has quit IRC22:59
*** salv-orlando has quit IRC23:14
*** salv-orlando has joined #openstack-neutron-ovn23:15
*** jckasper has joined #openstack-neutron-ovn23:16
openstackgerritMerged openstack/networking-ovn: Use compile_ovs() from Neutron tree
*** chandrav has quit IRC23:33

Generated by 2.14.0 by Marius Gedminas - find it at!