| *** fnaval has joined #openstack-lbaas | 00:15 | |
| *** xgerman has quit IRC | 00:27 | |
| *** sbfox has joined #openstack-lbaas | 00:51 | |
| *** vivek-ebay has joined #openstack-lbaas | 01:50 | |
| *** woodster__ has quit IRC | 01:55 | |
| dougwig | mestery: https://github.com/dougwig/a10_lbaas_driver/tree/juno-forwarding/tests-noop | 02:03 |
|---|---|---|
| *** sbalukoff has quit IRC | 02:04 | |
| dougwig | apologies for the delay, had to make a non-a10 driver version. i was going to add haproxy, with traffic passing testing, but i still can't get a pool to create. blogan and/or ctracey, that test shows the session persistence issue. | 02:04 |
| dougwig | nosetests test_noop.py | 02:04 |
| ctracey | dougwig: thanks | 02:07 |
| ctracey | man if it werent for members | 02:08 |
| dougwig | i'm wagering my test is broken for members, since it's never gotten that far. :) | 02:09 |
| *** fnaval has joined #openstack-lbaas | 02:09 | |
| ctracey | pool:members is causing all kinds of gried | 02:14 |
| ctracey | grief | 02:14 |
| ctracey | with unit testing | 02:14 |
| dougwig | didn't members require a pool in v1? | 02:14 |
| ctracey | i believe they were top-level though, right? | 02:15 |
| dougwig | ahh, i see. | 02:15 |
| dougwig | a trailblazer, you are. | 02:15 |
| ctracey | the nesting breaks a LOT of the existing plumbing | 02:15 |
| ctracey | i have all the tests except for members at this poit | 02:16 |
| ctracey | point( | 02:16 |
| ctracey | grrr | 02:16 |
| ctracey | that and testr sucks for telling you WTF failed | 02:16 |
| ctracey | "ConnectionFailed: Connection to neutron failed: Maximum attempts reached" | 02:16 |
| ctracey | makes perfect sense for a unit test, no? | 02:16 |
| ctracey | :) | 02:16 |
| dougwig | any idea where that damn session persistence is getting injected? i'm not even passing it in. | 02:17 |
| dougwig | compared to the messages when an import fails (which is 10 screens of nonsense characters), that's a downright sunny error. | 02:17 |
| ctracey | dougwig: that is what I was fighting for 2 days | 02:18 |
| ctracey | i have no looked at the session_persistence one yet | 02:18 |
| ctracey | but my units pass... | 02:18 |
| dougwig | when that hits, run it in 2.6, and the import error is printed normally. | 02:18 |
| ctracey | suspect something on the API side | 02:18 |
| ctracey | when i was working through that I didnt have the 26 interpretter installed | 02:19 |
| ctracey | now i do | 02:19 |
| ctracey | for a while I was only getting: | 02:19 |
| ctracey | https://www.irccloud.com/pastebin/4IDtFCMa | 02:20 |
| ctracey | nothing more | 02:20 |
| ctracey | which is super lovely | 02:20 |
| ctracey | that an no pdb, etc | 02:20 |
| dougwig | strace -f -f -s 1024, can shed some light sometimes. | 02:21 |
| dougwig | (prepend that before tester) | 02:21 |
| ctracey | hah | 02:21 |
| ctracey | i bet...though seriously...wtf? | 02:22 |
| dougwig | i have never met a test harness so eager to hide data, that's for sure. :) | 02:22 |
| ctracey | what about neutronclient itself? :) | 02:23 |
| ctracey | had a new guy run 'neutron quota-delete' | 02:23 |
| ctracey | guess what that does? :) | 02:23 |
| ctracey | warn you? no | 02:23 |
| ctracey | just deletes the current tenant's quotas | 02:23 |
| ctracey | the whole way that quotas are handled in both nova and neutron leaves a lot to be desired | 02:24 |
| dougwig | as barbican told us, only idiots need a warning. | 02:24 |
| dougwig | :) | 02:24 |
| ctracey | hah | 02:25 |
| ctracey | nova and neutron also want tenant-id's for quotas | 02:25 |
| ctracey | but will accept names | 02:25 |
| ctracey | cant tell you how many customers have tripped on this | 02:25 |
| dougwig | havana was super hit and miss for id vs name, but icehouse is pretty good about accepting names, if they're unique. | 02:26 |
| ctracey | i think i will add a sanity check..."does it look like a UUID? no? bail out" | 02:26 |
| ctracey | i think names are generally bad | 02:26 |
| ctracey | especially when customers are running with admin context | 02:26 |
| dougwig | not when i'm typing. nothing beats "nova start s1" | 02:27 |
| ctracey | oh i bet | 02:27 |
| ctracey | but if you ever use something like Cloud Foundry...all bets are off | 02:27 |
| dougwig | if i'm using a paas, then wtf am i typing openstack commands directly? | 02:34 |
| ctracey | dougwig: youre not... | 02:46 |
| ctracey | the issue is that things like sec groups can be specified as names...ie. default | 02:47 |
| ctracey | "default" can be used as a name across multiple tenancies | 02:47 |
| ctracey | the first person to run as an admin will break | 02:48 |
| dougwig | LOL | 02:49 |
| dougwig | oh, i can totally see that. | 02:50 |
| *** vivek-eb_ has joined #openstack-lbaas | 03:33 | |
| *** mestery_ has joined #openstack-lbaas | 03:34 | |
| *** mestery has quit IRC | 03:36 | |
| *** vivek-ebay has quit IRC | 03:37 | |
| *** vivek-eb_ has quit IRC | 03:38 | |
| *** mestery_ has quit IRC | 03:46 | |
| blogan | ctracey: found out the session_persistence issue, this is what the client is sending as session_persistence | 03:55 |
| blogan | "session_persistence ": null | 03:55 |
| blogan | that space after the e is causing the issue | 03:56 |
| *** mestery has joined #openstack-lbaas | 03:56 | |
| ctracey | blogan: thanks | 03:57 |
| ctracey | very odd as I dont do anything in that code | 03:58 |
| ctracey | https://www.irccloud.com/pastebin/pmih0klr | 03:58 |
| ctracey | and... | 03:58 |
| ctracey | https://www.irccloud.com/pastebin/4UlRogiU | 03:59 |
| *** mestery_ has joined #openstack-lbaas | 03:59 | |
| ctracey | i will take a look tomorrow | 03:59 |
| blogan | im sure its something along the lines of "if this attr name ends with an e, lets add a space at the end of it, because why not?" | 04:00 |
| blogan | i hope its not doing that, im being sarcastic of course | 04:00 |
| blogan | ctracey: thanks for looking into it, sorry i didn't just pool down the client code sooner | 04:01 |
| blogan | pool = pull | 04:01 |
| *** mestery has quit IRC | 04:02 | |
| *** woodster__ has joined #openstack-lbaas | 04:03 | |
| *** fnaval has quit IRC | 04:05 | |
| *** fnaval_ has joined #openstack-lbaas | 04:07 | |
| blogan | ctracey: just fyi, i added session_persistence to the actual cli pool create call and got this | 04:08 |
| blogan | http://pastebin.com/rQWk0spK | 04:08 |
| blogan | just more info for whenever you look into it | 04:08 |
| *** fnaval_ has quit IRC | 04:09 | |
| *** fnaval_ has joined #openstack-lbaas | 04:10 | |
| *** fnaval has joined #openstack-lbaas | 04:11 | |
| *** mestery_ has quit IRC | 04:11 | |
| *** fnaval_ has quit IRC | 04:14 | |
| dougwig | blogan: in the meantime, do you have an api client level test script we can provide for kyle? | 04:15 |
| blogan | dougwig: ive just been using a browser rest client | 04:16 |
| dougwig | got a log you can send me, and i'll script it quick? | 04:16 |
| dougwig | or else i can put something together in the morning. | 04:16 |
| blogan | well this thing stores my saved requests in json so let me see how parseable it is | 04:17 |
| blogan | hell i could whip up some curl requests fairly quickly | 04:17 |
| dougwig | that's all i was gonna do, too. | 04:18 |
| blogan | i'll do that right now | 04:19 |
| blogan | doesn't hurt to put them up on that deployment wiki page for now | 04:19 |
| dougwig | ok, awesome | 04:19 |
| dougwig | especially not if it gets the review train moving | 04:19 |
| blogan | i'm afraid to touch that first review, oleg gave it a +2 | 04:19 |
| blogan | i can't remember if hitting the rebase button wipes it all out, im pretty sure it does | 04:20 |
| dougwig | now you know how i felt with the no-op review. i overwrote so many +2's that I wanted to cry. | 04:20 |
| blogan | lol | 04:20 |
| dougwig | rebase doesn't wipe out -1's. | 04:20 |
| blogan | so the opposite of what i want | 04:20 |
| dougwig | meaning, rebase has different behavior than a push. no clue if +2 is in that mix. | 04:21 |
| *** mestery has joined #openstack-lbaas | 04:21 | |
| blogan | using the debug output of the client makes the curl thing much easier | 04:22 |
| blogan | yeah not worth testing it out in this case | 04:23 |
| blogan | the rebase i mean | 04:23 |
| *** sbfox has quit IRC | 04:26 | |
| *** sbfox has joined #openstack-lbaas | 04:33 | |
| *** fnaval has quit IRC | 04:45 | |
| blogan | dougwig: when you get a chance added the "scripts" to https://wiki.openstack.org/wiki/Neutron/LBaaS/DeployWithDevstack | 04:45 |
| *** HenryG is now known as HenryG_afk | 04:45 | |
| *** fnaval has joined #openstack-lbaas | 04:45 | |
| *** fnaval has quit IRC | 04:50 | |
| dougwig | testing now | 04:51 |
| dougwig | sub this for auth token insert: | 04:54 |
| dougwig | keystone token-get --wrap 0 | grep ' id ' | awk '{print $4}' | 04:54 |
| blogan | ahh | 04:54 |
| dougwig | one sec, folding that into the script that runs the whole matrix. | 04:54 |
| blogan | updated the page | 04:58 |
| dougwig | ctracey: your pool arg for listeners doesn't take names (which i know you hate, but goes against the rest of the cli.) | 05:15 |
| dougwig | hang on, might just be a bad error. | 05:17 |
| *** sbfox has quit IRC | 05:40 | |
| dougwig | mestery: test script updated with client workaround. grab those two files, enable the haproxy driver, it should create all the objects. might need to tweak the neutron network names in local_env. | 05:46 |
| dougwig | blogan: pythonified; | 05:47 |
| dougwig | https://www.irccloud.com/pastebin/cDIRqFz0 | 05:47 |
| *** ctracey has quit IRC | 06:01 | |
| *** woodster__ has quit IRC | 06:05 | |
| *** ctracey has joined #openstack-lbaas | 06:07 | |
| *** ctracey has quit IRC | 06:22 | |
| *** ctracey has joined #openstack-lbaas | 06:29 | |
| *** evgenyf has joined #openstack-lbaas | 07:23 | |
| *** sbalukoff has joined #openstack-lbaas | 08:03 | |
| *** evgenyf has quit IRC | 09:24 | |
| *** evgenyf has joined #openstack-lbaas | 09:37 | |
| *** enikanorov_ has joined #openstack-lbaas | 09:45 | |
| *** enikanorov has quit IRC | 09:46 | |
| *** vivek-ebay has joined #openstack-lbaas | 10:16 | |
| *** vivek-ebay has quit IRC | 10:20 | |
| *** HenryG_afk is now known as HenryG | 10:53 | |
| *** [1]evgenyf has joined #openstack-lbaas | 11:10 | |
| *** evgenyf has quit IRC | 11:13 | |
| *** [1]evgenyf is now known as evgenyf | 11:13 | |
| *** woodster__ has joined #openstack-lbaas | 12:32 | |
| mestery | dougwig: thanks sir! | 12:47 |
| *** sballe has joined #openstack-lbaas | 12:55 | |
| *** fnaval has joined #openstack-lbaas | 13:18 | |
| *** vivek-ebay has joined #openstack-lbaas | 13:23 | |
| *** vivek-ebay has quit IRC | 13:31 | |
| *** fnaval has quit IRC | 13:57 | |
| *** fnaval has joined #openstack-lbaas | 13:58 | |
| *** fnaval has quit IRC | 14:02 | |
| *** rolledback has joined #openstack-lbaas | 14:07 | |
| *** rolledback has quit IRC | 14:13 | |
| *** rolledback has joined #openstack-lbaas | 14:13 | |
| *** fnaval has joined #openstack-lbaas | 14:20 | |
| *** rolledback has quit IRC | 14:46 | |
| *** mlavalle has joined #openstack-lbaas | 15:04 | |
| rm_work | sbalukoff: what's your schedule look like today? pretty busy, or do you have some time? we've got some devs that actually have open hours this sprint and we're considering beginning some design discussions around Octavia -- and I know you guys will definitely need to be included there (and possibly already have some design drafts?) | 15:16 |
| rm_work | don't need to do it NOW, just letting you know (in fact, probably we'll be distracted for the next couple of hours anyway) | 15:17 |
| rm_work | I believe sballe also would want to be in any discussions there | 15:18 |
| rm_work | not sure who else | 15:18 |
| rm_work | first discussions might even just be figuring out *how to have our discussions* :P | 15:18 |
| sballe | rm_work, yes please and xgerman too | 15:23 |
| sballe | rm_work, We have a draft of a proposal. sbalukoff and I can go over it | 15:24 |
| sballe | I have better time tomorow. Today is wall to wall meetings. | 15:25 |
| rm_work | kk | 15:25 |
| rm_work | any chance we could get your proposal so we could at least read over it today? | 15:26 |
| rm_work | or would you rather go through it online | 15:26 |
| sballe | rm_work, sure. as soon as sbalukoff is online I am assuming he can send it to you | 15:29 |
| *** sbfox has joined #openstack-lbaas | 15:31 | |
| *** rolledback has joined #openstack-lbaas | 15:38 | |
| *** fnaval has quit IRC | 15:41 | |
| *** fnaval has joined #openstack-lbaas | 15:42 | |
| *** xgerman has joined #openstack-lbaas | 15:43 | |
| *** fnaval has quit IRC | 15:46 | |
| *** HenryG_ has joined #openstack-lbaas | 15:48 | |
| sballe | rm_work, blogan, dougwig What the status of the new API and Object model? is everythign merged and done? | 15:51 |
| *** HenryG has quit IRC | 15:52 | |
| *** HenryG_ is now known as HenryG_afk | 15:52 | |
| *** HenryG_afk is now known as HenryG | 15:52 | |
| dougwig | sballe: in gerrit, some rough edges, at least two of us have it creating objects successfully at this point. | 15:53 |
| sballe | dougwig, thanks :) | 15:54 |
| dougwig | blogan: you got double-rebased last night. | 15:58 |
| dougwig | and from the look of jenkins, not in a happy way. | 15:59 |
| dougwig | mestery: i expect the code reviews are messed up right now, due to the dependent rebasing. | 15:59 |
| *** evgenyf has quit IRC | 16:09 | |
| blogan | dougwig: i noticed :( | 16:11 |
| blogan | well time to spend some time on this | 16:12 |
| *** markmcclain has joined #openstack-lbaas | 16:15 | |
| mestery | blogan dougwig: And I was just about to try to re-setup my devstack and try things out. :) | 16:16 |
| dougwig | mestery: definitely WAIT. :) | 16:18 |
| dougwig | unless you like pain. i won't judge. | 16:18 |
| mestery | dougwig: Heh, ok, waiting | 16:18 |
| dougwig | wait, did i just put in effort to get a core to *not* do a review? that's a switch. :) | 16:19 |
| *** rolledback has quit IRC | 16:28 | |
| blogan | dougwig: it was for the betterment of mestery's sanity | 16:29 |
| *** rolledback has joined #openstack-lbaas | 16:29 | |
| dougwig | ahh, here comes the march of the CI emails. | 16:37 |
| *** rolledback has quit IRC | 16:39 | |
| *** crc32 has joined #openstack-lbaas | 16:41 | |
| *** sbfox has quit IRC | 16:42 | |
| *** dlundquist has joined #openstack-lbaas | 16:42 | |
| *** rolledback has joined #openstack-lbaas | 16:43 | |
| *** crc32 has quit IRC | 16:50 | |
| blogan | dougwig: well they'll come but may as well make the nit changes while im at it so this will take longer than me just pushing up the fixes | 16:50 |
| *** xgerman has quit IRC | 16:51 | |
| *** xgerman_ has joined #openstack-lbaas | 16:52 | |
| *** fnaval has joined #openstack-lbaas | 16:52 | |
| blogan | comments without spaces after the # are all over the neutron code, but i'll still fix mine anyway | 16:54 |
| *** rolledback has quit IRC | 16:55 | |
| *** sbfox has joined #openstack-lbaas | 16:57 | |
| *** dlundquist has quit IRC | 16:58 | |
| *** jorgem has joined #openstack-lbaas | 17:03 | |
| dougwig | that's technically a pep8 violation, but it might not be a hacking violation (or might have that rule ignored in neutron's tox file) | 17:04 |
| *** jorgem1 has joined #openstack-lbaas | 17:09 | |
| *** jorgem has quit IRC | 17:10 | |
| *** dlundquist has joined #openstack-lbaas | 17:11 | |
| *** fnaval has quit IRC | 17:13 | |
| *** fnaval has joined #openstack-lbaas | 17:13 | |
| *** rolledback has joined #openstack-lbaas | 17:15 | |
| *** crc32 has joined #openstack-lbaas | 17:17 | |
| *** fnaval has quit IRC | 17:18 | |
| blogan | mestery: things have been cleared up with the reviews | 17:25 |
| blogan | mestery: you're free to pound on it | 17:25 |
| *** fnaval has joined #openstack-lbaas | 17:26 | |
| *** sbfox1 has joined #openstack-lbaas | 17:35 | |
| mestery | blogan: thank you sir! | 17:35 |
| *** sballe has quit IRC | 17:36 | |
| *** sbfox has quit IRC | 17:37 | |
| dougwig | mestery: i'd suggest waiting for jenkins to finish first. | 17:38 |
| dougwig | blogan: you inspired me to finally put in mail filters. | 17:38 |
| mestery | dougwig: ++ | 17:39 |
| dougwig | commit message question, if anyone knows. for "Implements blueprint", do you use the blueprint filename after that (minus the .rst), or the name of the topic branch the bp got submitted under? | 17:40 |
| blogan | dougwig: whatever launchpad's blueprint name is | 17:57 |
| dougwig | ahh, three sources of truth. nice. | 18:00 |
| blogan | well thats why they say name the rst file the same name as the launchpad blueprint | 18:03 |
| blogan | but the reason I say launchpad blueprint name, is because when you click on that link it goes to launchpad | 18:04 |
| dlundquist | blogan: I might have hit a snag on the haproxy proxy driver member subnets: it seems we need to track which ports are associated with which load balancer some how. Although I'm not seeing how the existing haproxy driver does this | 18:10 |
| blogan | dlundquist: you mean for the member subnets? | 18:10 |
| dlundquist | yes | 18:10 |
| dougwig | wouldn't you either store the port with the member, or use the member ip+subnet to lookup the port? | 18:10 |
| blogan | dougwig: correct on your second part | 18:11 |
| blogan | dougwig: at least that is what is done in v1 with teh pool subnet in other drivers | 18:11 |
| dougwig | the first was a future enhancement for ease, not today. | 18:11 |
| blogan | ironically, the haproxy driver in v1 does not actually do anything with the pool subnet, so its only one arm loadbalancing | 18:11 |
| dlundquist | I'm wondering how we know which ports belong to the load balancer after deleting a member | 18:12 |
| *** markmcclain has quit IRC | 18:12 | |
| dougwig | serious question: why do you care? | 18:12 |
| dlundquist | if the name space is still present its easy find them, but if the node is rebooted we could leak ports | 18:12 |
| dougwig | don't those ports belong to the nova instance, and it's an instance cleanup problem? | 18:13 |
| dlundquist | no, the haproxy namespace needs to plug into unique member subnet | 18:14 |
| dlundquist | in case there is no L3 path from the loadbalancer port to the member | 18:14 |
| blogan | dlundquist is right | 18:14 |
| dlundquist | It doesn't look like the existing HAproxy driver does this | 18:14 |
| blogan | it does not | 18:15 |
| dougwig | i thought the suggestion was to use static routes to the subnet gateway? | 18:15 |
| dlundquist | How does that work if a member is on an isolated network? | 18:15 |
| *** markmcclain has joined #openstack-lbaas | 18:16 | |
| sbalukoff | dlundquist: Are you perhaps suggesting a supplemental table to keep track of this is going to be necessary (not one exposed by the API of course)? | 18:16 |
| dlundquist | I think it the load balancer port mappings need to be persisted in the DB somewhere | 18:17 |
| *** markmcclain1 has joined #openstack-lbaas | 18:17 | |
| dougwig | i think if the v1 driver doesn't support this, it might be out of scope for the ref driver. | 18:17 |
| dougwig | and isolated how? don't all neutron subnets have a gateway you can lookup, assuming service account permissions? | 18:18 |
| sbalukoff | The v1 driver is supposed to support multiple member subnets. | 18:18 |
| dougwig | right, via routes. i remember this exact conversation at the meetup. | 18:18 |
| dlundquist | dougwig: you can create a subnet without a gateway | 18:18 |
| blogan | the v1 driver is supposed to support one pool subnet | 18:19 |
| dougwig | so are we fixing a v1 bug, or fixing it on the sly into v2? can we phase it as a separate commit, and not bundling it all into what we're trying to get in? | 18:19 |
| blogan | however it does not, i talked to mark about it and he was under the assumption it did but he thinks maybe it got removed | 18:19 |
| blogan | dougwig: I've actually got some code for it but was going to bring it up on rather it should be included in this or later since v1 didn't have it working | 18:20 |
| dlundquist | I'm fine leaving this deficiency, but it does seem to be a deficiency in the v1 HAProxy driver | 18:20 |
| *** markmcclain has quit IRC | 18:21 | |
| dougwig | concur that it's a bug. i don't know, we're already at risk, the lure of slipping in fixes can be the death of a timeline. | 18:21 |
| blogan | yes and I'd rather not put it in | 18:21 |
| blogan | because of it adding extra things people need to look at | 18:21 |
| sbalukoff | Aah. | 18:21 |
| dlundquist | sounds good | 18:21 |
| blogan | however, not many are actually looking at that code right now | 18:21 |
| sbalukoff | Well, I'm OK with that, too. | 18:22 |
| dougwig | or if you can structure it as a dependent review behind the core ref driver; i don't want to lose the work. i also don't want it to distract or add risk. | 18:22 |
| blogan | oh god, antoher dependent review? | 18:22 |
| dougwig | lol. | 18:22 |
| dougwig | don't you have a script to rollback the magic yet? | 18:22 |
| blogan | mo' dependent reviews, mo' problems | 18:22 |
| blogan | i need to reply to that ML thread about this and just tell people to look at the dependent commits' hashes and if its not the same as in gerrit, something bad happened | 18:23 |
| dougwig | i said that in my reply. | 18:23 |
| dougwig | it was on the list of "how to change your stuff later." | 18:23 |
| blogan | oh i guess i had my dougwig filter enabled | 18:23 |
| dougwig | it's a common filter. | 18:24 |
| blogan | more useful than the gerrit CI filter i have | 18:24 |
| * dougwig pouts. | 18:25 | |
| dlundquist | Should I file a bug against the existing v1 HAProxy driver? | 18:27 |
| dlundquist | Should we bubble up an unsupported configuration exception when building a load balancer with member subnet_id set? | 18:28 |
| dougwig | assuming that they'll just use an ip without a subnet if it's routable? | 18:29 |
| dlundquist | yes, if no subnet is specified it should assume the IP is reachable via the load balancers default route | 18:31 |
| dougwig | at the driver level, sounds reasonable. at the plugin level, not so much. | 18:32 |
| blogan | what sounds reasonable at the driver level? | 18:32 |
| dlundquist | I think surfacing an error we encounter an supported configuration is better than not implementing the configuration completely | 18:47 |
| *** sbfox1 has quit IRC | 19:03 | |
| *** sbfox has joined #openstack-lbaas | 19:03 | |
| *** rolledback has quit IRC | 19:12 | |
| *** sbfox has quit IRC | 19:17 | |
| blogan | dlundquist: are you saying if a user creates a member with subnet_id bubble up ane xception? | 19:32 |
| blogan | mestery: all those reviews have passed the CI's, except the last one, which was expected | 19:32 |
| mestery | blogan: nice nice nice! | 19:33 |
| *** openstackgerrit has quit IRC | 19:48 | |
| dlundquist | blogan: Yes, if the user creates a member with the subnet_id set, the HAProxy driver should raise an exception since that feature is not supported by the driver. | 19:58 |
| *** markmcclain1 has quit IRC | 19:59 | |
| *** markmcclain has joined #openstack-lbaas | 20:00 | |
| blogan | dlundquist: ah okay I see what you mean now. I'm on the fence with that, but will do what the majority wants. My reason for being on the fence is that it didn't work in v1 and it didn't throw that exception | 20:11 |
| blogan | and we intend on fixing it after all is said and done | 20:12 |
| dlundquist | okay, should we atleast log that were not implementing that feature | 20:25 |
| *** sbfox has joined #openstack-lbaas | 20:43 | |
| *** clev has joined #openstack-lbaas | 20:47 | |
| *** banix has joined #openstack-lbaas | 20:50 | |
| *** sbfox has quit IRC | 20:54 | |
| *** rolledback has joined #openstack-lbaas | 20:55 | |
| banix | was asking a question from blogan on the neutron channel and he mentioned there is a lbaas channel as well :) so switching to here | 20:56 |
| *** openstackgerrit has joined #openstack-lbaas | 20:58 | |
| dougwig | banix: the answer is that you get status of members from the haproxy driver via the stats() API. | 20:58 |
| dougwig | there's a super recent ML thread about it, where someone on heat tested it out and posted an example. | 20:59 |
| banix | my question was about how the status of pool members get updated by health monitors. I am wondering if it would be possible for having some kind of notification issued when the status of a pool member gets updated. I do not see the current code doing anything like that. Would that be reasonable? | 20:59 |
| *** rolledback has quit IRC | 21:00 | |
| banix | dougwig: thanks. yes i am trying to get to the next question about the notification; was wondering what you think of such a thing | 21:00 |
| *** dlundquist has left #openstack-lbaas | 21:04 | |
| *** dlundquist has joined #openstack-lbaas | 21:10 | |
| *** fnaval has quit IRC | 21:12 | |
| *** fnaval has joined #openstack-lbaas | 21:13 | |
| blogan | banix: you basically want notifications sent when the state of a member changes | 21:14 |
| banix | blogan: yes, and here is the why if you want to know: | 21:15 |
| blogan | oh i know why, its very helpful for automated systems and just any operators to get notified, isntead of manually checking the API | 21:16 |
| banix | blogan: for consumption by Heath for scaling groups | 21:16 |
| blogan | Heath? you mean Heat? | 21:17 |
| banix | blogan: exactly but this is a more specific use case | 21:17 |
| banix | yeas Heat | 21:17 |
| *** fnaval has quit IRC | 21:17 | |
| blogan | ah okay, making sure there wasn't another project i'm not aware of | 21:17 |
| blogan | so what kind of notification would Heat need and where would it need it? | 21:19 |
| blogan | i understand Heat from a very high level, haven't dealt too much with it | 21:19 |
| *** jorgem1 has quit IRC | 21:20 | |
| banix | blogan: just to make sure I am not getting ahead of myself here, there is a proposal in that community (and there has been discussions about it for some time) for scaling group to be able to detect changes in the scaling group | 21:21 |
| banix | blogan: so right now the scaling group does not do the health maintenance of members as AWS does | 21:23 |
| blogan | i haven't read those discussions, so excuse my ignorance, but out of curiousity what will happen when a member sends a notification that is down? | 21:23 |
| blogan | banix: yeah and lbaas is probably further behind AWS than what the scaling group is | 21:25 |
| banix | blogan: can lead to starting new instances (deploying new resources) for example | 21:26 |
| blogan | ok thats what i figured | 21:26 |
| blogan | but its more of an action a user can define | 21:26 |
| banix | blogan: yes | 21:26 |
| blogan | banix: well lbaas is in a bit of a transition in Juno, and it will bleed over into Kilo as well | 21:26 |
| banix | blogan: i think this is just one piece to get to a better scaling group… | 21:27 |
| banix | blogan: yes, i follow the good work of the lbaas team and see the patches and the ml discussions | 21:27 |
| banix | blogan: i wanted to first get a sense of what you think of adding something like that … | 21:28 |
| dougwig | i think it'd be good to add, fwiw. | 21:30 |
| blogan | banix: well i'm definitely all for it, I know it won't be bale to be done in Kilo, and I'm unsure about it in Kilo | 21:30 |
| blogan | but yes it should be done | 21:30 |
| banix | blogan: you mean not in Juno | 21:30 |
| banix | blogan: maybe Kilo? | 21:30 |
| blogan | oh yeah sorry | 21:30 |
| dougwig | blogan has been non-stop coding, has a new kid, and his wife had her tonsils out. how he's even responding at all, i don't know. | 21:31 |
| blogan | banix: i would say there's a chance | 21:31 |
| banix | dougwig: blogan: thanks for answering my questions | 21:31 |
| banix | yeah with all that is going on in Neutron it has been a really busy cycle | 21:31 |
| banix | and we seem to have a lot more to land for the basic stuff we need in the next month or so; so i understand | 21:32 |
| blogan | banix: no problem, soryr if I couldn't give you a definitive answer of if it can get into Kilo or not | 21:33 |
| banix | blogan: I understand. thanks for the input. so I know if we want to have a chance in Kilo we have to be very pro active and start ealy :) | 21:34 |
| banix | dougwig: just noticed the comment about thre new kid! so congratulations to blogan | 21:35 |
| banix | blogan: you must be feeling exhausted but things will get better hang in there :) | 21:37 |
| blogan | banix: thanks a lot! | 21:37 |
| blogan | banix: thats what everyone tells me, I haven't had an uninterrupted night's sleep in forever | 21:37 |
| *** rolledback has joined #openstack-lbaas | 21:38 | |
| banix | blogan: i know; may i ask how old your child is now? | 21:38 |
| blogan | banix: oh he's 9 months now | 21:39 |
| blogan | banix: so he's not exactly new, but he's still new to me | 21:39 |
| blogan | he's about to start walking so thats going to be even more exhausting always having to watch him like a hawk | 21:40 |
| *** sbfox has joined #openstack-lbaas | 21:41 | |
| banix | blogan: All the best to him and your family! you will get more sleep in a few months! yes indeed; walking and crawling at that age when they essentially don’t have an idea about the dangers is really tiring for the parents; you have to be watchful all every second | 21:41 |
| blogan | banix: thanks, best to you too and your family (sounds like you've got your own little one(s)) | 21:42 |
| banix | blogan: thanks I have a son. and as i said, as they get older the connection and communication gets much stronger. Anyhow I better let you go :) | 21:44 |
| banix | By the way, I have ben reviewing the lbaas code and have had some minor comments (mainly style wise and some questions). I am Mohamad Banikazemi on the review board. Thanks for responding quickly. | 21:45 |
| blogan | banix: yep looking forward to it. take care of yourself. don't let us forget about the member notification feature you want. I'll try to remember it too | 21:45 |
| banix | blogan: will do. thanks. | 21:45 |
| blogan | banix: ahh okay, you caught taht one TODO: i forgot to fix | 21:46 |
| blogan | hopefully it was just one | 21:46 |
| banix | just a nit. very very minor; i think the code is in a very good shape. | 21:47 |
| blogan | banix: i may end up fixing that in the next commit just so I don't have to add another patchset to that review | 21:48 |
| *** rolledback has quit IRC | 21:48 | |
| blogan | banix: would that work for you? | 21:48 |
| banix | blogan: yes of course. | 21:48 |
| blogan | this chain of reviews, and chain branches has been less than ideal so the further up the chain a change happens the greater the chance something bad will happen | 21:49 |
| blogan | and actually now a change that happens at the bottom of the chain can somehow manifest itself into the top piece | 21:50 |
| *** banix has quit IRC | 22:09 | |
| *** HenryG is now known as HenryG_afk | 22:18 | |
| *** sbfox has quit IRC | 22:19 | |
| *** sbfox has joined #openstack-lbaas | 22:20 | |
| dougwig | blogan secretly loves it when you rewrite all of his commits with older code. it makes him feel loved. | 22:43 |
| *** fnaval has joined #openstack-lbaas | 22:46 | |
| * blogan goes into the fetal position and cries | 22:48 | |
| blogan | that reminds me, i want to check something | 22:49 |
| *** fnaval has quit IRC | 23:06 | |
| *** fnaval has joined #openstack-lbaas | 23:07 | |
| *** fnaval has quit IRC | 23:11 | |
| *** dlundquist has left #openstack-lbaas | 23:13 | |
| *** fnaval has joined #openstack-lbaas | 23:15 | |
| *** banix has joined #openstack-lbaas | 23:23 | |
| *** banix has quit IRC | 23:37 | |
| *** banix has joined #openstack-lbaas | 23:41 | |
| *** HenryG_afk is now known as HenryG | 23:45 | |
| *** banix has quit IRC | 23:46 | |
| *** banix has joined #openstack-lbaas | 23:55 | |
| *** banix has quit IRC | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!