| *** woodster__ has quit IRC | 01:15 | |
| *** fnaval has quit IRC | 01:28 | |
| *** fnaval has joined #openstack-lbaas | 01:29 | |
| *** fnaval has quit IRC | 01:33 | |
| *** mestery has joined #openstack-lbaas | 01:42 | |
| *** mlavalle has quit IRC | 02:43 | |
| *** woodster__ has joined #openstack-lbaas | 02:52 | |
| ctracey | dougwig: you saying that session_persistence is wrong in client or API? | 04:09 |
|---|---|---|
| dougwig | blogan didn't use any underscores. and I didn't pass it at all. | 04:10 |
| ctracey | it does use underscores unless that has recently changed | 04:13 |
| ctracey | dougwig: https://github.com/oslbaas/neutron/blob/bp/lbaas-api-and-objmodel-improvement-haproxy-driver/neutron/extensions/loadbalancerv2.py#L227 | 04:15 |
| ctracey | but i am confused as to why it is not sticking on the backend | 04:15 |
| ctracey | even when I pass it | 04:15 |
| blogan | hello | 05:18 |
| dougwig | heya | 05:19 |
| dougwig | ctracey: sec, looking into it on my end. | 05:20 |
| blogan | dougwig: what are saying about the plugin holding off telling the drivers about the listeners? | 05:21 |
| dougwig | just that it looks like when the "parent" gets created, the downhill notification is auto-magic because of how the foreign keys are setup. | 05:22 |
| dougwig | assuming any model change triggers a driver call, that is. | 05:22 |
| dougwig | haven't gotten things running end-to-end enough to test it. | 05:22 |
| blogan | well the plugin still does a check on all non loadbalancer CUD calls to see if it is linked to a loadbalancer, and if it is it sends it to the driver | 05:23 |
| blogan | is that what you mean? | 05:25 |
| dougwig | sort of; but it's a second order issue for me right now. still trying to get things up enough to pass to the drivers at all. | 05:26 |
| dougwig | this should go on the todo list somewhere: | 05:26 |
| dougwig | blogan: for 105610, can you swap LOADBALANCERV2 for LOADBALANCER, for the noop driver, in etc/neutron.conf? | 05:27 |
| blogan | yeah i can do htat, i got another change going into the one before that, but im holding off until tomorrow | 05:27 |
| blogan | ill update both in the same chain rebase | 05:28 |
| blogan | you get over that migration issue about lbaas_listeners being missing? | 05:28 |
| blogan | oh yeah step in devstack | 05:28 |
| dougwig | yes, i somehow had stale crap in my neutron db, which is odd, considering it was a fresh devstack. | 05:28 |
| blogan | should checkout the review first, then run devstack | 05:28 |
| dougwig | i edited the wiki with all the tweaks i had to make. | 05:29 |
| blogan | did you put in that you should check hte review out first then run devstack? | 05:29 |
| blogan | of cours ethat would assume neutron was there first | 05:30 |
| blogan | oh yeah ctracey: api is session_persistence | 05:32 |
| dougwig | blogan: i got it working the other way around, actually. well, "working". | 05:33 |
| blogan | neutron-db-manage? | 05:33 |
| dougwig | the db is correct, neutron server is running the new bits and the drivers are initializing. | 05:33 |
| dougwig | first drop tables, then db-manage, yes. | 05:33 |
| dougwig | i just had the cli problems above, and hadn't done anything straight to the api yet. | 05:34 |
| blogan | hmm | 05:38 |
| *** blogan_ has joined #openstack-lbaas | 05:49 | |
| blogan_ | can't recreate what the pastebin showed the cli was returning | 06:01 |
| blogan_ | im hitting the sack, see ya tomorrow | 06:07 |
| *** blogan_ has quit IRC | 06:07 | |
| *** woodster__ has quit IRC | 07:35 | |
| *** enikanorov__ has quit IRC | 09:52 | |
| *** evgenyf has joined #openstack-lbaas | 10:45 | |
| *** HenryG_afk is now known as HenryG | 12:12 | |
| *** woodster__ has joined #openstack-lbaas | 12:52 | |
| *** samuelbercovici has joined #openstack-lbaas | 12:58 | |
| *** fnaval has joined #openstack-lbaas | 13:06 | |
| *** sballe has joined #openstack-lbaas | 13:54 | |
| *** samuelbercovici1 has joined #openstack-lbaas | 13:58 | |
| *** samuelbercovici has quit IRC | 14:01 | |
| *** samuelbercovici1 is now known as samuelbercovici | 14:01 | |
| *** fnaval has quit IRC | 14:10 | |
| *** rolledback has joined #openstack-lbaas | 14:13 | |
| *** samuelbercovici1 has joined #openstack-lbaas | 14:16 | |
| *** samuelbercovici has quit IRC | 14:19 | |
| *** samuelbercovici1 is now known as samuelbercovici | 14:19 | |
| *** fnaval has joined #openstack-lbaas | 14:36 | |
| sballe | Morning | 15:02 |
| *** openstackgerrit has joined #openstack-lbaas | 15:02 | |
| *** sbfox has joined #openstack-lbaas | 15:06 | |
| *** sbalukoff has quit IRC | 15:07 | |
| *** xgerman has joined #openstack-lbaas | 15:12 | |
| *** jorgem has joined #openstack-lbaas | 15:40 | |
| *** sbfox has quit IRC | 15:44 | |
| *** TrevorV_ has joined #openstack-lbaas | 15:45 | |
| *** TrevorV_ has quit IRC | 15:46 | |
| blogan | hello | 15:53 |
| blogan | all | 15:53 |
| *** mestery has quit IRC | 15:55 | |
| TrevorV | hi | 16:03 |
| TrevorV | blogan | 16:03 |
| TrevorV | how | 16:03 |
| TrevorV | goes | 16:03 |
| TrevorV | it | 16:03 |
| blogan | hi TrevorV | 16:03 |
| blogan | who are you? | 16:03 |
| *** evgenyf has quit IRC | 16:05 | |
| *** markmcclain has joined #openstack-lbaas | 16:17 | |
| blogan | markmcclain: do you know if mestery will be on today? | 16:22 |
| markmcclain | blogan: he should be back today | 16:23 |
| blogan | markmcclain: or if you can shed some light on a question, we have two specs that need to be accepted into juno-3, but they haven't yet, they're the tls and l7 specs. I think mestery will be up to accepting them today if we can get it in today. | 16:24 |
| blogan | markmcclain: would you agree? | 16:24 |
| markmcclain | blogan: let me check the list | 16:25 |
| markmcclain | l7 should get an approval | 16:26 |
| *** mlavalle has joined #openstack-lbaas | 16:26 | |
| blogan | markmcclain: that woudl be great, but I think tls is probably just as, if not more, important | 16:26 |
| markmcclain | tls is tied to barbican | 16:27 |
| markmcclain | so that's causing a bit of project sync issue which is why it's on the fence | 16:27 |
| samuelbercovici | hi all | 16:28 |
| samuelbercovici | markmcclain: what is the "project sync" issue at hand? | 16:29 |
| blogan | sameulbercovici: i'm assuming its because what our tls spec requires from barbican ismerged in yet iether | 16:31 |
| markmcclain | it's more bureaucratic that barbican is on track for graduation | 16:31 |
| samuelbercovici | markmcclain: the bp can be approved with the pre-condition that if barbican will not be graduated, this will be delayed | 16:32 |
| samuelbercovici | right? | 16:32 |
| markmcclain | yeah it's more finding space for it… concerned that the v2+L7 is a huge backlog to clear | 16:34 |
| markmcclain | I think the promise to delay seems acceptable | 16:34 |
| markmcclain | if the other work gets backed up | 16:35 |
| samuelbercovici | sore. | 16:35 |
| samuelbercovici | sure | 16:35 |
| samuelbercovici | approving the BP just gives us the green light to work on this. | 16:36 |
| samuelbercovici | with the hope that we can make it to the release | 16:36 |
| blogan | +1 sam | 16:36 |
| jorgem | Adam told me that the Barbican work was implemented and pending review. | 16:37 |
| markmcclain | jorgem: go to know | 16:37 |
| jorgem | He's out today but I'm trying to get more details from him | 16:37 |
| *** markmcclain has quit IRC | 16:38 | |
| jorgem | I scared mark away :( | 16:38 |
| jorgem | lol | 16:39 |
| blogan | way to go jorge | 16:39 |
| samuelbercovici | LOL | 16:39 |
| jorgem | Here is the work Adam completed ==> https://review.openstack.org/#/c/107845/ | 16:40 |
| samuelbercovici | this is nice work. it is not even mandatory thought. | 16:41 |
| jorgem | Which item is mandatory? | 16:42 |
| *** markmcclain has joined #openstack-lbaas | 16:42 | |
| jorgem | I don't think any of them are from what Adam & Carlos told me. Thus TLS is not blocked. | 16:42 |
| jorgem | There work was meant to make it easier. | 16:42 |
| samuelbercovici | +1 | 16:43 |
| samuelbercovici | the only work which may be needed, not sure where it will reside is the libraries carlos is doing | 16:44 |
| samuelbercovici | hi mark | 16:45 |
| *** evgenyf has joined #openstack-lbaas | 16:56 | |
| dougwig | blogan: you're not seeing the two errors i hit? (not the non-default stuff; i can work around that.) | 16:58 |
| blogan | dougwig: which two errors? | 16:59 |
| blogan | session_persistence? | 16:59 |
| blogan | brb | 16:59 |
| dougwig | these two: | 16:59 |
| dougwig | https://www.irccloud.com/pastebin/FP493i5F | 17:00 |
| dougwig | hey folks, last chance to give +1's to blueprints you really want: https://etherpad.openstack.org/p/lbaas_reviews | 17:02 |
| dougwig | (possibly past last chance, but hey.) | 17:02 |
| *** rolledback has quit IRC | 17:10 | |
| *** samuelbercovici has quit IRC | 17:21 | |
| *** evgenyf has quit IRC | 17:22 | |
| dougwig | blogan, i assume you renamed something? | 17:23 |
| dougwig | ./neutron/tests/unit/services/loadbalancer/drivers/a10networks/test_thunder_driver.py:19:1: H302 import only modules.'from neutron.tests.unit.db.loadbalancer import test_db_loadbalancerv2' does not import a module | 17:23 |
| dougwig | nope, looks like my rebased repo is messed up. | 17:24 |
| *** barclaac|2 has quit IRC | 17:29 | |
| *** mestery has joined #openstack-lbaas | 17:33 | |
| *** mestery has quit IRC | 17:33 | |
| *** mestery has joined #openstack-lbaas | 17:34 | |
| blogan | ah taht session_persistence is probably broken in the haproxy driver | 17:42 |
| blogan | mestery: can we get these 3 specs approved today: https://review.openstack.org/#/c/98640/ https://review.openstack.org/#/c/99709/ https://review.openstack.org/#/c/100931/ | 17:45 |
| mestery | blogan: I'll see what I can do. Given these are all LBaaS specific, I'm inclined to approve them post-SAD (today), and if they land, they land. | 17:47 |
| mestery | blogan: Makes sense? | 17:47 |
| blogan | you mean if the code is completed and reviewed in time then it would be able to land? | 17:47 |
| blogan | mestery: ^^ | 17:48 |
| mestery | blogan: Correct | 17:49 |
| *** sbfox has joined #openstack-lbaas | 17:49 | |
| blogan | mestery: are there major issues with approving the spec but, in the case of tls, if barbican is not ready then don't merge the code in? | 17:50 |
| blogan | mestery: or is it just more the volume of changes that lbaas is getting that you'd rather just take the wait and see approach? | 17:51 |
| mestery | blogan: I think a little of both to be honest | 17:51 |
| blogan | mestery: ok understood, would you mind if i sent out an email to the lbaas group quoting you on if the code is fully implemented and fully reviewed then it can go into Juno? | 17:53 |
| blogan | mainly beacuse people may freak out if they see it didnt get accepted in time | 17:53 |
| blogan | the specs that is | 17:53 |
| mestery | blogan: Yes, though for TLC, the barbican question is slightly open still | 17:54 |
| dougwig | for tls & l7, we also need to get the driver changes/object change info out to the vendor driver authors before the last possible second, or those two items will *only* be in the ref driver. | 17:54 |
| blogan | mestery: ok i'll make a point to mention that, is it because of the uncertainty that barbican will have everything needed for tls? or some other issues we are unaware of? | 17:54 |
| mestery | blogan: The main uncertaintiy is around if barbican will graduate or not, I think that's it. | 17:55 |
| blogan | mestery: ah okay thanks, ill mention that. appreciate your patience | 17:55 |
| dougwig | blogan: one sec on that email. | 17:57 |
| blogan | dougwig: im starting to think we may not have enough time to the older drivers updated with tls and l7, that woudl be the shim that woudl do that right? | 17:57 |
| blogan | dougwig: okay | 17:57 |
| dougwig | mestery: what's the last day for code submissions for juno-3? | 17:57 |
| mestery | blogan: https://wiki.openstack.org/wiki/Juno_Release_Schedule | 17:57 |
| mestery | blogan: Sept 4 if code is already submitted | 17:58 |
| *** fnaval has quit IRC | 17:58 | |
| dougwig | aug 21 is code not submitted? | 17:58 |
| *** sbalukoff has joined #openstack-lbaas | 17:59 | |
| dougwig | if | 17:59 |
| blogan | mestery: i'm kind of reiterating jorge's question here, but july 20 was the last day for the specs to be accepted, where is that date specified? | 17:59 |
| mestery | blogan: That was in an email which I sent, I can dig it out. | 18:00 |
| dougwig | blogan: i think we need like an august 14th deadline or something for the tls/l7 ref changes submitted, then. we've got 3 vendors at present writing new drivers, that will need at least a little time to respond in kind. | 18:00 |
| mestery | blogan: I think it was 2-3 weeks ago. | 18:01 |
| mestery | blogan: SPD/SAD is something which nova and neutron took part in. | 18:01 |
| mestery | To try and control the spec rush near the end. | 18:01 |
| dougwig | it was mentioned at every neutron and lbaas irc meeting for the last several weeks. | 18:01 |
| blogan | mestery: no dont dig it out I was just wondering if it was wikified | 18:01 |
| blogan | dougwig: obviously im just ignorant | 18:02 |
| blogan | been hard to keep up with everything | 18:02 |
| dougwig | ha, just saying that i haven't seen it on any official release wikis. i just heard about it at meetings. | 18:02 |
| blogan | i shall get better though! | 18:02 |
| blogan | once the rush dies down | 18:02 |
| blogan | one does SPD and SAD stand for? | 18:03 |
| blogan | nvm found it | 18:04 |
| blogan | spec proposal deadline, spec approval deadlilne | 18:04 |
| dougwig | openstack does like its acronyms. | 18:05 |
| blogan | dougwig: still hold of on sending that email? | 18:07 |
| dougwig | sorry, i sent this as food for thought on deadlines for the email: "blogan: i think we need like an august 14th deadline or something for the tls/l7 ref changes submitted, then. we've got 3 vendors at present writing new drivers, that will need at least a little time to respond in kind." | 18:07 |
| dougwig | or we need to somehow communicate the object changes before the ref driver is submitted. | 18:08 |
| blogan | dougwig: are they writing new drivers only, or are they also going to implement tls adn l7? | 18:08 |
| dougwig | is there a difference? | 18:09 |
| dougwig | as in, i don't know, but we should give them a fighting chance. | 18:09 |
| blogan | well there is because they will probably want to base their cdoe off a review that hast he tls adn l7 object model changes vs only the v2 lbaas | 18:10 |
| *** rolledback has joined #openstack-lbaas | 18:11 | |
| dougwig | personally, i expect to submit two reviews in the juno-3 timeframe; the first based off of v2, the second with additions based on l7/tls. i'm not sure it's wise for anyone to wait until the latter review is ready. | 18:11 |
| blogan | what im getting as is i want to be clear about what exactly needs to go in by august 14th: just new driver interface implementation of their drivers, or that and tls and l7 | 18:11 |
| *** rolledback has quit IRC | 18:11 | |
| dougwig | i think we just need to let them know what the interface will be; i don't care if it's pre-release, or WIP, or a google doc. | 18:11 |
| *** rolledback has joined #openstack-lbaas | 18:12 | |
| blogan | okay well im pretty sure we have a consensus on the L7 and tls stuff so that can easily be derived | 18:12 |
| dougwig | i agree, just want to make sure we prioritize the interface changes, because several people have to serialize around that point. | 18:12 |
| blogan | okay so is it agreed that vendors can't just implement l7 and tls in their old drivers and expect the shim to do some kind of translation so their old drivers can still do tls/l7, which I'm pretty sure wouldn't even be possible so its probably a moot comment | 18:13 |
| blogan | or question | 18:13 |
| blogan | just thinking about how todo that hurts my brain | 18:14 |
| dougwig | oh hell no. | 18:18 |
| dougwig | :) | 18:18 |
| dougwig | i hadn't even ever considered it. | 18:18 |
| dougwig | damnit, i think i screwed up your gerrit commits. and i'm not even sure how, except that a git rebase on a gerrit review branch must've auto-propagated (and seemingly not in the way that it looks in my tree.) | 18:19 |
| dougwig | can you look? | 18:19 |
| *** sbfox has quit IRC | 18:20 | |
| xgerman | dougwig it seems your last review pulled some constant changes | 18:20 |
| dougwig | if you look here: https://review.openstack.org/#/c/105331/13/neutron/plugins/common/constants.py,unified | 18:20 |
| dougwig | i definitely undid some constants changes. in a way that's opposite to how the merge looks locally. | 18:21 |
| xgerman | thanks, | 18:21 |
| xgerman | Michael was asking... | 18:22 |
| blogan | lol dougwig what did you do | 18:22 |
| dougwig | pull brandon's n-1 changesets. | 18:22 |
| *** johnsom has joined #openstack-lbaas | 18:22 | |
| dougwig | blogan: i don't even know! | 18:22 |
| blogan | what did you intend on pushing up? | 18:22 |
| dougwig | i added a line to neutron.conf, ran 'git review', it said some things about auto-rebasing dependencies, magic happened. | 18:23 |
| dougwig | bad, bad magic. | 18:23 |
| blogan | lol | 18:23 |
| blogan | i was going to add that in the other one i thought | 18:24 |
| dougwig | this was avendor specific one. | 18:24 |
| dougwig | a vendor | 18:24 |
| blogan | ah okay | 18:24 |
| dougwig | ok, what's the best way to rollback? | 18:24 |
| blogan | i can do it | 18:25 |
| dougwig | ok, thanks. | 18:25 |
| blogan | well if you want to go ahead | 18:25 |
| blogan | but i can do it quickly, and add your change if you want | 18:26 |
| dougwig | please, go for it. i think i'd have to setup your four branches and submit each separately, so it'd take me a bit. i managed to screw up five changesets with one branch. | 18:26 |
| blogan | lol | 18:26 |
| blogan | well i was planning on pushing code up to this one anyway | 18:27 |
| blogan | so it would have caused the chain reaction | 18:27 |
| blogan | what change did you need to make though? | 18:27 |
| TrevorV | Is anyone else having issues using dev-stack right now? My neutron is failing with a DB issue. | 18:28 |
| dougwig | if you pulled in the last 20 minutes, it's broken. standby. | 18:29 |
| dougwig | if not, please paste the error. | 18:29 |
| TrevorV | I pulled before I went to lunch. Like, over an hour ago | 18:29 |
| dougwig | is the error by chance about a missing lbaas.listeners table? | 18:30 |
| TrevorV | Detected added foreign key for column 'id' on table u'ml2_brocadeports' | 18:30 |
| TrevorV | that's what mine was | 18:30 |
| TrevorV | orrr... maybe that's just an INFO and not the actual failure | 18:31 |
| TrevorV | Hold on | 18:31 |
| TrevorV | dougwig, blogan says to give him a minute and I'll pull and see if I still have a problem | 18:33 |
| blogan | xgerman: do you know what michael johnson ws referencing with his comment : "Looks good. One note about including a gerrit comment in the code."? | 18:38 |
| johnsom | blogan: Hi. I think Gerrit ate my comment | 18:40 |
| blogan | lol i was looking for mjohnson in the userlist, i forgot you were johnsom | 18:41 |
| blogan | gerrit does like to eat comments | 18:41 |
| xgerman | hungry gerrit? especially around lunch time? | 18:42 |
| blogan | lol dougwig you managed to change the topic as well | 18:43 |
| dougwig | fuck me. | 18:43 |
| blogan | gerrit already did that | 18:43 |
| johnsom | There was a discussion about the exceptions and the extension loading magic. | 18:44 |
| ptoohill | lol | 18:44 |
| johnsom | Its the extension loading magic that happens with neutron. Even though they are separate REST urls, the extension loading stores the resources in a dictionary with the keys being the resource names. So with two resources named 'pools', it does a dictionary.update() if the resource already exists. Tough for me to explain but the code is here: | 18:44 |
| johnsom | to quote | 18:44 |
| *** evgenyf has joined #openstack-lbaas | 18:45 | |
| blogan | ah yeah i remember typing that up, but i still dont understand what you mean gerrit comment though? | 18:45 |
| blogan | you mean my code comments that mention to remove it once the v1 is totally gone? | 18:45 |
| *** TrevorV_ has joined #openstack-lbaas | 18:46 | |
| johnsom | I didn't see those comments in the code. I just thought it would be a good idea to include the reason in code comments as it could come up later. | 18:48 |
| blogan | johnsom: ahh, i didn't put comments on that particular exception that oleg was commenting about but I did later in the extension definition where I made the extension define the attributes that were nto required and also in the plugin code where it raises that particular exceptions | 18:50 |
| *** mestery has quit IRC | 18:51 | |
| johnsom | Ah, ok | 18:52 |
| *** mestery_ has joined #openstack-lbaas | 18:56 | |
| *** mestery_ has quit IRC | 18:57 | |
| *** mestery has joined #openstack-lbaas | 18:57 | |
| *** sbfox has joined #openstack-lbaas | 18:58 | |
| *** sbfox has quit IRC | 19:15 | |
| *** enikanorov_ has joined #openstack-lbaas | 19:16 | |
| mestery | blogan: I just went +2 on all 3 BPs, waiting for markmcclain to also +2 and then he or I can +A them. | 19:21 |
| mestery | blogan: Thanks for your and all the LBaaS team members work in this area! | 19:21 |
| blogan | mestery: the specs? | 19:23 |
| mestery | blogan: Correct, the specs | 19:23 |
| blogan | mestery: awesome, thanks a bunch on that! | 19:24 |
| mestery | blogan: no problem sir! | 19:25 |
| dougwig | mestery: thank you! | 19:25 |
| mestery | dougwig: sure! | 19:26 |
| *** sbfox has joined #openstack-lbaas | 19:34 | |
| dougwig | ok, git is happy again. many thanks, blogan | 19:35 |
| *** evgenyf has quit IRC | 19:38 | |
| *** fnaval has joined #openstack-lbaas | 19:42 | |
| blogan | dougwig: lets wait to see if gerrit is happy | 19:43 |
| *** evgenyf has joined #openstack-lbaas | 19:44 | |
| sbalukoff | mestery: That's great! Thanks! | 20:05 |
| *** Youcef has joined #openstack-lbaas | 20:06 | |
| *** rolledback has quit IRC | 20:15 | |
| *** evgenyf has quit IRC | 20:15 | |
| *** rolledback has joined #openstack-lbaas | 20:19 | |
| *** sbfox has quit IRC | 20:28 | |
| *** sbfox has joined #openstack-lbaas | 20:34 | |
| *** rolledback has quit IRC | 20:43 | |
| *** jorgem has quit IRC | 21:01 | |
| *** fnaval has quit IRC | 21:12 | |
| *** fnaval has joined #openstack-lbaas | 21:22 | |
| *** TrevorV_ has quit IRC | 21:29 | |
| *** mestery has quit IRC | 21:33 | |
| *** dlundquist has joined #openstack-lbaas | 21:35 | |
| sbalukoff | markmcclain: Will you have time today to look at the specs that Kyle has +2'ed to see whether y'all are comfortable merging these? | 22:00 |
| *** fnaval_ has joined #openstack-lbaas | 22:00 | |
| *** fnaval has quit IRC | 22:01 | |
| *** fnaval_ has quit IRC | 22:05 | |
| *** fnaval has joined #openstack-lbaas | 22:06 | |
| markmcclain | sbalukoff: | 22:11 |
| markmcclain | sbalukoff: yes | 22:11 |
| mlavalle | blogan: when re-starting the neutron server (q-svc) I get the folowwing: 2014-07-21 22:13:56.898 15583 TRACE neutron dirs = self._retrieve_deployed_instance_dirs() | 22:15 |
| mlavalle | 2014-07-21 22:13:56.898 15583 TRACE neutron File "/opt/stack/neutron/neutron/services/loadbalancer/drivers/haproxy/nonagent_namespace_driver.py", line 144, in _retrieve_deployed_instance_dirs | 22:15 |
| mlavalle | 2014-07-21 22:13:56.898 15583 TRACE neutron return [dir for dir in os.listdir(self.state_path) | 22:15 |
| mlavalle | 2014-07-21 22:13:56.898 15583 TRACE neutron OSError: [Errno 2] No such file or directory: '/opt/stack/data/neutron/lbaas/v2' | 22:15 |
| blogan | mlavalle: ahh fresh system bug catch | 22:16 |
| mlavalle | blogan: I am glad I caught my first lbaas v2 bug :-) | 22:16 |
| blogan | mlavalle: ha first of many i'm sure | 22:16 |
| blogan | mlavalle: to be fair this is a WIP review | 22:17 |
| blogan | mlavalle: to get around this you can either create that directory on your system or use the loggingnoop driver | 22:17 |
| mlavalle | blogan: just create the directory empty? | 22:18 |
| blogan | mlavalle: yep | 22:18 |
| *** dlundquist has quit IRC | 22:18 | |
| mlavalle | blogan: ok…. one other thing…. in the instrctions in the wiki, you indicate to stop the neutron service in the screen utility…. It would be clearer if you state that the screen name is q-svc | 22:19 |
| mlavalle | blogan: actually, if you are ok with it, I can fix it | 22:20 |
| blogan | mlavalle: sure you can fix it, dougwig added that so I will blame him again | 22:20 |
| blogan | feels great shifting blame to other people, i'd be a good politician | 22:20 |
| mlavalle | blogan: ok, I blame dougwig as well, so it's official | 22:21 |
| * dougwig is soaking in in the blame. | 22:21 | |
| dougwig | i'm not sure i edited the part in question, but i'll take the blame anyway. | 22:22 |
| blogan | lol | 22:22 |
| *** fnaval_ has joined #openstack-lbaas | 22:22 | |
| *** fnaval has quit IRC | 22:25 | |
| mlavalle | blogan: the v2 in /opt/stack/data/neutron/lbaas/v2 is also a directory? | 22:25 |
| blogan | mlavalle: that is correct | 22:25 |
| *** dlundquist has joined #openstack-lbaas | 22:26 | |
| sbalukoff | markmcclain: Thanks! | 22:26 |
| xgerman | sbalukoff here is another doc review for you https://review.openstack.org/#/c/108517/ | 22:26 |
| blogan | xgerman: is that the official api docs for the v2 lbaas api? | 22:27 |
| xgerman | sbalukoff +1 | 22:27 |
| xgerman | blogan, yes. Min is creating them | 22:27 |
| blogan | xgerman: is she just using libra as a central place to store it or is this part of libra? | 22:28 |
| sbalukoff | xgerman: I'll have a look, eh! | 22:28 |
| xgerman | sorry, I got confused | 22:28 |
| xgerman | that might be the one libfra change I asked her to look at | 22:29 |
| xgerman | too many parallel conversations | 22:29 |
| blogan | xgerman: still hung over from germany's world cup win? | 22:29 |
| mlavalle | blogan: next problem: 2014-07-21 22:26:27.070 18958 TRACE neutron ProgrammingError: (ProgrammingError) (1146, "Table 'neutron_ml2.lbaas_loadbalancers' doesn't exist") 'SELECT lbaas_loadbalancers.tenant_id AS lbaas_loadbalancers_tenant_id, lbaas_loadbalancers.id AS lbaas_loadbalancers_id, lbaas_loadbalancers.name AS lbaas_loadbalancers_name, lbaas_loadbalancers.description AS lbaas_loadbalancers_description, | 22:29 |
| mlavalle | lbaas_loadbalancers.vip_subnet_id AS lbaas_loadbalancers_vip_subnet_id, lbaas_loadbalancers.vip_port_id AS lbaas_loadbalancers_vip_port_id, lbaas_loadbalancers.vip_address AS lbaas_loadbalancers_vip_address, lbaas_loadbalancers.status AS lbaas_loadbalancers_status, lbaas_loadbalancers.admin_state_up AS lbaas_loadbalancers_admin_state_up, providerresourceassociations_1.provider_name AS providerresourceassociations_1_provider_name | 22:29 |
| mlavalle | providerresourceassociations_1.resource_id AS providerresourceassociations_1_resource_id \nFROM lbaas_loadbalancers LEFT OUTER JOIN providerresourceassociations AS providerresourceassociations_1 ON lbaas_loadbalancers.id = providerresourceassociations_1.resource_id' () | 22:29 |
| blogan | mlavalle: i knew that would happen | 22:30 |
| blogan | mlavalle: restack | 22:30 |
| xgerman | bloga, yep. I asked my kids every day "Who is world champion?" :-) | 22:30 |
| blogan | mlavalle: or there is a neutron-db-manage command | 22:30 |
| mlavalle | blogan: just type "neutron-db-manage" | 22:31 |
| dougwig | mlavalle: it'll be: "neutron-db-manage --config-file /etc/neutron/neutron.conf --config-file /etc/neutron/plugins/ml2/ml2_conf.ini upgrade head" | 22:33 |
| blogan | dougwig: i believe that is correct | 22:33 |
| blogan | should add taht in the wiki too | 22:33 |
| blogan | i didn't test the wiki out step-by-step, my bad!! | 22:34 |
| mlavalle | dougwig: and after that restart neutron-server, right? | 22:34 |
| dougwig | yes. | 22:35 |
| blogan | yes | 22:35 |
| * mlavalle trying | 22:35 | |
| dougwig | (ideally neutron-server isn't running during the migration.) | 22:35 |
| mlavalle | dougwig: yeah…. neutron server is down at this moment | 22:35 |
| mlavalle | dougwig, blogan: from neutron-db-manage sqlalchemy.exc.OperationalError: (OperationalError) (1050, "Table 'lbaas_healthmonitors' already exists") "\nCREATE TABLE lbaas_healthmonitors (\n\ttenant_id VARCHAR(255), \n\tid VARCHAR(36) NOT NULL, \n\ttype ENUM('PING','TCP','HTTP','HTTPS') NOT NULL, \n\tdelay INTEGER NOT NULL, \n\ttimeout INTEGER NOT NULL, \n\tmax_retries INTEGER NOT NULL, \n\thttp_method VARCHAR(16), \n\turl_path | 22:37 |
| mlavalle | VARCHAR(255), \n\texpected_codes VARCHAR(64), \n\tstatus VARCHAR(16) NOT NULL, \n\tadmin_state_up BOOL NOT NULL, \n\tPRIMARY KEY (id), \n\tCHECK (admin_state_up IN (0, 1))\n)\n\n" () | 22:37 |
| markmcclain | I've +2'd the TLS specs | 22:37 |
| markmcclain | had comments on the L7 spec… I like blogan's suggestion of nested relationship | 22:38 |
| blogan | markmcclain: yeah! though I don't think it will be easy with the extension resource loader | 22:39 |
| dougwig | mlavalle: stale db; open with mysql (mysql -u root neutron_ml2), and drop all the existing lbaas tables. then re-migrate. | 22:41 |
| * mlavalle trying | 22:42 | |
| blogan | mlavalle: yeah migrations won't handle that gracefully | 22:43 |
| blogan | i debated using drop if exists but no one else did in previous migrations so I didn't, just so the down migration would work | 22:43 |
| sbalukoff | markmcclain and blogan: If holding off for a nested relationship model means we don't get L7 switching until Kilo, then I would advocate going for the relatively simple L7 policy / rule structure outlined in the spec. | 22:45 |
| sbalukoff | Also, the spec essentially lays out what we'd talked about on the mailing list for several months. It's a little frustrating to get told we need nested relationships now. :P | 22:46 |
| blogan | sbalukoff: blame me, I was too focused on other things to really focus on those specs | 22:46 |
| sbalukoff | Also, in my experience, it's extremely rare to need advanced logic in an L7 rule. :P Something like 95% of the policies we see in production consist of 1 rule looking for a hostname or URL path. | 22:47 |
| sbalukoff | Most application developers perfer to do advanced application logic in the application layer, not the load balancer. | 22:48 |
| blogan | so you're saying no need for L7? great! | 22:49 |
| blogan | kidding | 22:49 |
| sbalukoff | blogan: Haha! I'mma come break your knees. | 22:49 |
| dougwig | sbalukoff: i have a pile of counter-examples, but i expect you're correct for the consumer level stuff. | 22:49 |
| sbalukoff | Look, if y'all think we can get nested relationships in time for Juno (and by that, I mean, convince the people representing vendors whose products probably won't support this anyway), then that's great... but I'd really rather not delay until Kilo for layer 7 switching. | 22:51 |
| sbalukoff | Would it be a terrible liability to support L7 switching as suggested in the spec? | 22:51 |
| sbalukoff | (Would this prevent us from using nested relationships in a version 2 of this design?) | 22:51 |
| blogan | isn't this a version 2 already though? | 22:52 |
| blogan | and changing teh API contract is a huge deal | 22:52 |
| sbalukoff | blogan: Are you anticipating zero API changes between Juno and Kilo? | 22:52 |
| sbalukoff | I'm not... | 22:52 |
| blogan | more like API additions, not deprecations | 22:52 |
| blogan | adding thigns to an API is not breaking contract | 22:53 |
| sbalukoff | Oh! So, do you think this would entail a depreciation? | 22:53 |
| blogan | i assumed you meant add the nested relationship in K and deprecate the root l7policy and l7rules | 22:53 |
| sbalukoff | Would nested relationships do away with rool l7policy and l7rules? | 22:54 |
| sbalukoff | Really, I'm not sure how you envision nested relationships working. | 22:54 |
| blogan | honestly I would hope so, i wouldn't want two ways to do it | 22:54 |
| blogan | lol well i guess im not sure how you are envisioning it either | 22:55 |
| mlavalle | blogan, dougwig: I got the eutron server running. I have to go now, but I will run the tempest test tomorrow | 22:55 |
| blogan | mlavalle: okay cool, you will probably run into erros that happen in the driver so sorry, if that happens often enough try out the LoggingNoop driver | 22:56 |
| *** sbfox has quit IRC | 22:56 | |
| mlavalle | blogan: thanks, will do | 22:56 |
| *** fnaval_ has quit IRC | 22:56 | |
| sbalukoff | Oh wait! | 22:56 |
| sbalukoff | blogan: Are you talking about the URL path? | 22:57 |
| blogan | yes | 22:57 |
| blogan | what the hell are you talking about? | 22:57 |
| *** fnaval has joined #openstack-lbaas | 22:57 | |
| sbalukoff | Sorry... total brainfart. | 22:57 |
| sbalukoff | I thought this had to do with earlier comments about advanced logic. | 22:57 |
| sbalukoff | AND and OR rules, etc. | 22:58 |
| sbalukoff | Derp. | 22:58 |
| blogan | ha lol i was wondering why you were bringing that up | 22:58 |
| sbalukoff | Yeah... Ok, so what you're suggesting should be a "minor" change to the spec as is. | 22:58 |
| sbalukoff | And yes, I'm in favor of it. :) | 22:58 |
| blogan | oh look whos changed his tune! | 22:58 |
| sbalukoff | I'm as wiggly as a worm, eh. | 22:58 |
| sbalukoff | And yes-- I will totally change my tune if you show me how I'm being stupid. | 22:59 |
| *** fnaval_ has joined #openstack-lbaas | 22:59 | |
| blogan | there will still be a challenge in getting the neutron extension loading working with 3rd level resources | 22:59 |
| sbalukoff | (Which isn't that hard, most times.) | 22:59 |
| blogan | i bet i can work on that and get it working | 22:59 |
| sbalukoff | If we have to resort to a fall-back way of doing this which doesn't exactly match the spec, is that a show-stopper? | 22:59 |
| xgerman | I bet - you are the miracle worker here | 22:59 |
| sbalukoff | Yes, indeed! | 23:00 |
| blogan | well the fall-back way is having them as root level objects | 23:00 |
| blogan | orrrr | 23:00 |
| sbalukoff | Or l7rule being a child of l7policy | 23:00 |
| sbalukoff | Which isn't a bad way to go about it. | 23:00 |
| *** mlavalle has quit IRC | 23:00 | |
| xgerman | +1 | 23:01 |
| sbalukoff | Well, crapsticks, then! What do we need to change about this spec to get that approved? | 23:01 |
| blogan | since 2nd level subresources work, keep l7policy as a subresource of listener, but then have the l7rules bet be an attribute of l7policy, still store it in its own table but dont represent it in the url | 23:01 |
| *** fnaval has quit IRC | 23:01 | |
| sbalukoff | blogan: And since all the rules are ANDed together, order doesn't matter... yeah, that could work. | 23:02 |
| xgerman | well, mestery is busy: https://twitter.com/mestery/status/491353727463194625/photo/1 | 23:02 |
| sbalukoff | Doing a beta-test of that UDP load balancing session, eh? | 23:02 |
| xgerman | yep, 4 years OpenStack in MN | 23:03 |
| sbalukoff | Heh! The Seattle celebration is this Thursday. | 23:03 |
| sbalukoff | I won't be there, because friends got me tickets to Book of Mormon... but it ought to be fun for those who do go. | 23:03 |
| blogan | ah man i missed that when it was here in san antonio! | 23:04 |
| xgerman | yep, I am in San Diego and we don't have much OpenStack | 23:04 |
| blogan | i was talking abotu book or mormon | 23:04 |
| blogan | of | 23:04 |
| blogan | sbalukoff: even if the L7 spec doesn't get accepted today, it'll probably get accepted soon. Plus mestery told me that even al 3 of those didn't get accepted and the code landed in gerrit and was fully approved he'd still accept it in Juno | 23:06 |
| *** markmcclain has quit IRC | 23:06 | |
| sbalukoff | Oh, heck yes. | 23:06 |
| blogan | sbalukoff: so don't worry | 23:07 |
| blogan | about that at least | 23:07 |
| blogan | I am worried about the shear amount of volume we have to do | 23:07 |
| sbalukoff | Yes, looking at this spec, there are other problems, too-- like the all caps constants suggestion you made in revision 17. | 23:07 |
| xgerman | yeah, you got a good bragian | 23:07 |
| sbalukoff | Damn! I'm late for a meeting. BBIAB. | 23:07 |
| xgerman | well, i would go soft | 23:07 |
| xgerman | on the all caps | 23:07 |
| xgerman | but yeah, if they need to change by all means | 23:08 |
| sbalukoff | NOT ME, DAMMIT! ;) | 23:08 |
| blogan | alright im headed home | 23:08 |
| blogan | take it easy! | 23:08 |
| xgerman | see ya | 23:09 |
| *** dlundquist has quit IRC | 23:12 | |
| *** dlundquist has joined #openstack-lbaas | 23:26 | |
| *** fnaval_ has quit IRC | 23:51 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!