| sbalukoff | Sorry, was in a meeting. | 00:01 |
|---|---|---|
| sbalukoff | Put Barbican on the left and Keystone on the right. | 00:01 |
| *** jroyall has joined #openstack-lbaas | 00:01 | |
| *** jroyall is now known as joeroyall | 00:01 | |
| sbalukoff | (We're not reading manga here, eh. ;) ) | 00:02 |
| *** mlavalle has quit IRC | 00:05 | |
| *** sbfox1 has quit IRC | 00:08 | |
| *** xgerman has quit IRC | 00:14 | |
| rm_you | pff | 00:24 |
| *** vivek-ebay has joined #openstack-lbaas | 00:24 | |
| rm_you | sbalukoff: hadn't thought of that :P | 00:24 |
| rm_you | though I'm probably going to redo the whole thing as a websequencediagram | 00:24 |
| sbalukoff | dougwig: In light of the non-support of eventlet in python 3, should we kill this: https://review.openstack.org/#/c/118827/ | 00:29 |
| dougwig | commented. | 00:32 |
| *** vivek-ebay has quit IRC | 00:53 | |
| *** joeroyall has left #openstack-lbaas | 00:55 | |
| sbalukoff | Anyone interested, feel free to edit as you see fit: https://wiki.openstack.org/wiki/Octavia/Non-arbitrary_Decisions#What_should_we_call_the_back-end_VM_.2F_container_.2F_machine_.2F_appliance_.2F_thingy.3F | 01:07 |
| *** vivek-ebay has joined #openstack-lbaas | 01:37 | |
| sbalukoff | This is awesome, and I think we should use it: https://storyboard.openstack.org/#!/project/list | 01:43 |
| sbalukoff | (Just throwing that out there.) | 01:43 |
| *** sbfox has joined #openstack-lbaas | 01:52 | |
| *** mestery has quit IRC | 01:57 | |
| *** mestery_ has joined #openstack-lbaas | 01:57 | |
| *** sbalukoff has quit IRC | 02:10 | |
| *** sbfox has quit IRC | 02:12 | |
| *** openstackgerrit has quit IRC | 02:33 | |
| *** openstack has joined #openstack-lbaas | 03:41 | |
| *** sendak.freenode.net sets mode: +ns | 03:41 | |
| *** sendak.freenode.net sets mode: -o openstack | 03:46 | |
| -sendak.freenode.net- *** Notice -- TS for #openstack-lbaas changed from 1410234100 to 1403021244 | 03:46 | |
| *** sendak.freenode.net sets mode: +ct-s | 03:46 | |
| *** vivek-eb_ has joined #openstack-lbaas | 03:46 | |
| *** enikanorov__ has joined #openstack-lbaas | 03:46 | |
| *** mestery_ has joined #openstack-lbaas | 03:46 | |
| *** ptoohill has joined #openstack-lbaas | 03:46 | |
| *** nealph has joined #openstack-lbaas | 03:46 | |
| *** RaginBajin has joined #openstack-lbaas | 03:46 | |
| *** woodster_ has joined #openstack-lbaas | 03:46 | |
| *** rm_work has joined #openstack-lbaas | 03:46 | |
| *** amotoki has joined #openstack-lbaas | 03:46 | |
| *** masteinhauser has joined #openstack-lbaas | 03:46 | |
| *** enikanorov has joined #openstack-lbaas | 03:46 | |
| *** dkehnx1 has joined #openstack-lbaas | 03:46 | |
| *** dougwig has joined #openstack-lbaas | 03:46 | |
| *** jkoelker has joined #openstack-lbaas | 03:46 | |
| *** ctracey has joined #openstack-lbaas | 03:46 | |
| *** pckizer has joined #openstack-lbaas | 03:46 | |
| *** redrobot has joined #openstack-lbaas | 03:46 | |
| *** TrevorV has joined #openstack-lbaas | 03:46 | |
| *** rm_you has joined #openstack-lbaas | 03:46 | |
| *** johnsom_ has joined #openstack-lbaas | 03:46 | |
| *** blogan has joined #openstack-lbaas | 03:46 | |
| *** a2hill has joined #openstack-lbaas | 03:46 | |
| *** sendak.freenode.net changes topic to "https://wiki.openstack.org/wiki/Neutron/LBaaS" | 03:46 | |
| *** sbfox has joined #openstack-lbaas | 03:48 | |
| *** woodster_ has quit IRC | 03:55 | |
| *** jroyall has joined #openstack-lbaas | 03:59 | |
| *** jroyall is now known as joeroyall | 03:59 | |
| *** rm_work has quit IRC | 04:03 | |
| *** joeroyall has quit IRC | 04:04 | |
| *** HenryG has joined #openstack-lbaas | 04:07 | |
| *** rm_work has joined #openstack-lbaas | 04:08 | |
| *** rm_work is now known as rm_work|away | 04:08 | |
| blogan | sbalukoff: that storyboard looks much better than launchpad, but i think launchpad will suffice now as that storyboard doesn't look too mature | 04:17 |
| *** vivek-eb_ has quit IRC | 04:52 | |
| *** vivek-ebay has joined #openstack-lbaas | 04:53 | |
| *** sbalukoff has joined #openstack-lbaas | 05:24 | |
| *** vivek-ebay has quit IRC | 06:08 | |
| *** jroyall has joined #openstack-lbaas | 06:35 | |
| *** jroyall is now known as joeroyall | 06:35 | |
| *** joeroyall has quit IRC | 07:01 | |
| *** dkehnx1 has quit IRC | 07:36 | |
| *** dkehnx1 has joined #openstack-lbaas | 07:37 | |
| *** jschwarz has joined #openstack-lbaas | 07:57 | |
| *** sbfox has quit IRC | 08:21 | |
| *** sbalukoff has quit IRC | 08:23 | |
| *** ptoohill-oo has joined #openstack-lbaas | 09:15 | |
| *** ptoohill has quit IRC | 09:15 | |
| *** openstackgerrit has joined #openstack-lbaas | 09:20 | |
| *** amotoki has quit IRC | 11:57 | |
| *** ptoohill-oo has quit IRC | 12:31 | |
| *** ptoohill has joined #openstack-lbaas | 12:31 | |
| *** mestery_ is now known as mestery | 12:37 | |
| *** mestery has quit IRC | 13:17 | |
| *** mestery has joined #openstack-lbaas | 13:17 | |
| *** crc32 has joined #openstack-lbaas | 13:24 | |
| *** mestery has quit IRC | 13:34 | |
| *** vivek-ebay has joined #openstack-lbaas | 13:34 | |
| *** mestery has joined #openstack-lbaas | 13:34 | |
| *** markmcclain has joined #openstack-lbaas | 13:38 | |
| *** vivek-ebay has quit IRC | 14:07 | |
| *** crc32 has quit IRC | 14:15 | |
| *** sbfox has joined #openstack-lbaas | 14:43 | |
| *** TrevorV_ has joined #openstack-lbaas | 14:52 | |
| *** xgerman has joined #openstack-lbaas | 14:59 | |
| *** sbfox has quit IRC | 15:01 | |
| *** markmcclain has quit IRC | 15:05 | |
| *** mlavalle has joined #openstack-lbaas | 15:11 | |
| *** mlavalle has quit IRC | 15:12 | |
| *** mlavalle has joined #openstack-lbaas | 15:12 | |
| dougwig | morning | 15:17 |
| dougwig | is that storyboard what they're talking about replacing launchpad with? | 15:18 |
| xgerman | morning | 15:20 |
| xgerman | I welcome anything which replaces launchpad | 15:21 |
| xgerman | wonder why they can't get some free Jira like Apache | 15:21 |
| jschwarz | dougwig, ping | 15:24 |
| jschwarz | dougwig, Can you elaborate regarding "Ownership reply"? | 15:24 |
| dougwig | heya. | 15:24 |
| dougwig | just referring to the nobody question. | 15:25 |
| jschwarz | yes.. I didn't understand the comment | 15:25 |
| dougwig | running daemons as a real user is dangerous. especially one that can get root, like stack. | 15:25 |
| dougwig | far lesser evil to make that socket writable to all. | 15:25 |
| jschwarz | A discussion worth having | 15:25 |
| jschwarz | This patch is getting bigger and bigger and it looks like this proposal isn't strong enough and I'll have to drop it eventually | 15:26 |
| jschwarz | dougwig, btw lbaas-agent can get root (runs under stack) and we have no problem with that? | 15:26 |
| dougwig | i think i'm only suggesting a looser chmod on one side, and leaving the nobody on the other? that should shrink you by one line. | 15:27 |
| dougwig | lbaas-agent isn't internet facing. | 15:27 |
| *** mikedillion has joined #openstack-lbaas | 15:27 | |
| jschwarz | I see your point | 15:28 |
| dougwig | openstack definitely seems to subscribe to the soft and gooey center model of security, but haproxy isn't in the center. neutron-server, lbaas-agent, are. | 15:28 |
| jschwarz | I gotta get home which will take me about an hour... will you be here then? | 15:29 |
| dougwig | yes. | 15:29 |
| jschwarz | dougwig, thanks :) | 15:29 |
| dougwig | i'm just waking up. :) | 15:29 |
| jschwarz | 18:29:43 | 15:30 |
| jschwarz | :p | 15:30 |
| jschwarz | see you guys later | 15:30 |
| *** jschwarz has quit IRC | 15:30 | |
| *** vivek-ebay has joined #openstack-lbaas | 15:38 | |
| *** vivek-ebay has quit IRC | 15:41 | |
| *** woodster_ has joined #openstack-lbaas | 15:41 | |
| *** RaginBajin has quit IRC | 15:44 | |
| *** jroyall has joined #openstack-lbaas | 15:49 | |
| *** jroyall is now known as joeroyall | 15:49 | |
| *** RaginBajin has joined #openstack-lbaas | 15:56 | |
| *** RaginBajin has joined #openstack-lbaas | 15:57 | |
| xgerman | dougwig I gave blogan a +2; in case you like to +2 and merge | 16:18 |
| blogan | yes +2! | 16:20 |
| *** rm_work|away is now known as rm_work | 16:27 | |
| dougwig | quick -1 | 16:28 |
| blogan | no! | 16:28 |
| dougwig | i had to show my love. | 16:28 |
| blogan | i dont want your love | 16:28 |
| blogan | -1 bc you diagree with the name "amphora"? | 16:29 |
| blogan | disagree | 16:29 |
| dougwig | oh, someone is fishing for a -2. | 16:29 |
| johnsom_ | Wasn't being too picky on the TLS stuff as I assumed when it merged they would adjust as needed | 16:29 |
| blogan | johnsom_: definitely, kind of just followed what neutron lbaas had, but I don't expect we'll be implementing that at first | 16:30 |
| dougwig | quick question on why we're carrying forward the TERMINATED_HTTPS nonsense? can't we do https + cert here, since we're not carrying already shipped neutron baggage? | 16:30 |
| *** markmcclain has joined #openstack-lbaas | 16:30 | |
| blogan | oh yeah yall wanted just HTTP protocol and then TCP for HTTPS passhtrough, and then HTTPS would be termination? | 16:30 |
| blogan | obviously TCP not only for https passthrough | 16:31 |
| dougwig | right. | 16:31 |
| dougwig | i'd be fine with the review with just nuking that one line. | 16:31 |
| *** markmcclain has quit IRC | 16:32 | |
| xgerman | let's keep HTTPS -- there is some special stuff we can do as opposed to TCP | 16:32 |
| blogan | i still prefer just having the HTTPS protocol and if a user wants to terminate then they provide the tls_container_id | 16:32 |
| blogan | so I'm sensing this will be something we won't have a consensus on right away | 16:33 |
| *** sbfox has joined #openstack-lbaas | 16:33 | |
| dougwig | xgerman: we're not talking about removing it. just using TCP when it's really a complete pass-through, and using HTTPS when we can do the special stuff. or i'm fine with blogan's, where it's a pass-through until a cert is defined. | 16:33 |
| *** markmcclain has joined #openstack-lbaas | 16:33 | |
| *** jorgem has joined #openstack-lbaas | 16:34 | |
| blogan | so since I think we all agree we don't like having the TERMINATED_HTTPS protocol, I will remove it and that leaves TCP, HTTP, and HTTPS which are the protocols we need to accomplish what everyone wants | 16:34 |
| dougwig | +1 | 16:35 |
| *** jschwarz has joined #openstack-lbaas | 16:35 | |
| jschwarz | hola | 16:35 |
| dougwig | jschwarz: heya | 16:35 |
| *** crc32 has joined #openstack-lbaas | 16:36 | |
| jschwarz | dougwig, at the end of the day the issue with the permissions is the least of my worries here | 16:36 |
| xgerman | blogan +1 | 16:36 |
| jschwarz | dougwig, that whole loop-around because the haproxy don't want to write straight to disk is complicating my code | 16:37 |
| dougwig | jschwarz: i went through the rest, and will +1. also note that the stats file uses a 666 file to avoid the nobody issue. | 16:37 |
| jschwarz | dougwig, I don't think this is the right solution here | 16:37 |
| openstackgerrit | Brandon Logan proposed a change to stackforge/octavia: Initial migration for database structure https://review.openstack.org/114671 | 16:37 |
| jschwarz | brb | 16:37 |
| dougwig | jschwarz: it's going to be tough getting past the cores until kilo, no matter what. | 16:38 |
| blogan | dougwig, xgerman: if yall could +2 again and +A it if you feel it is complete enough | 16:38 |
| dougwig | T minus 22 minutes until hipster overload death. | 16:38 |
| jschwarz | dougwig, I lost hope getting anything before K starts anyway | 16:39 |
| jschwarz | dougwig, Another solution could be implemented in the deployment-tool level, ie. packstack will change the appropriate syslog settings so that will dump to file instead | 16:40 |
| dougwig | or forward to a remote. that is likely cleaner. | 16:41 |
| jschwarz | This design is better 'design-wise' but harder to implement | 16:41 |
| jschwarz | a remote? | 16:41 |
| dougwig | remote syslog. | 16:42 |
| jschwarz | aye | 16:42 |
| johnsom_ | Missed the running typo, but fine | 16:42 |
| jschwarz | the haproxy guys said that a standing solution they normally offer to clients is simply set-up multiple syslog daemons, one for each haproxy, and that pair will dump to files | 16:43 |
| *** joeroyall has quit IRC | 16:54 | |
| *** jschwarz has quit IRC | 16:55 | |
| *** jschwarz has joined #openstack-lbaas | 16:58 | |
| *** mestery has quit IRC | 16:59 | |
| *** mestery has joined #openstack-lbaas | 17:00 | |
| xgerman | blogan should merge, sorry for the delay... | 17:00 |
| *** vivek-ebay has joined #openstack-lbaas | 17:07 | |
| *** jschwarz has quit IRC | 17:16 | |
| *** jschwarz has joined #openstack-lbaas | 17:18 | |
| openstackgerrit | Trevor Vardeman proposed a change to stackforge/octavia: Initial creation of db models, modules, and tests https://review.openstack.org/116718 | 17:20 |
| *** jschwarz has quit IRC | 17:22 | |
| *** sbfox has quit IRC | 17:42 | |
| *** jschwarz has joined #openstack-lbaas | 17:42 | |
| openstackgerrit | A change was merged to stackforge/octavia: Initial migration for database structure https://review.openstack.org/114671 | 17:45 |
| *** sbfox has joined #openstack-lbaas | 17:53 | |
| rm_work | dougwig / xgerman / sbalukoff: http://goo.gl/x6oBZu | 18:10 |
| xgerman | ok | 18:10 |
| xgerman | lgtm -- so the trust stays for the live of the LB? | 18:12 |
| *** jschwarz_ has joined #openstack-lbaas | 18:13 | |
| *** jschwarz has quit IRC | 18:16 | |
| rm_work | the Trust stays *forever* | 18:18 |
| rm_work | I should actually indicate that | 18:18 |
| *** jschwarz_ has quit IRC | 18:20 | |
| xgerman | yep.. | 18:20 |
| *** crc32 has quit IRC | 18:20 | |
| blogan | should that be something we clean up after the load balancer is deleted? | 18:24 |
| rm_work | hmm | 18:26 |
| rm_work | The trust itself isn't really linked to the LB, it's just user_account<-->lbaas_account | 18:27 |
| rm_work | so we could only remove it if they had zero LBs left in the system | 18:27 |
| blogan | so if a user has multiple tls terminated load balancers then it wouldn't create another trust? | 18:27 |
| rm_work | or, zero TLS Terminations | 18:27 |
| blogan | one for both? | 18:27 |
| rm_work | yes | 18:27 |
| rm_work | no additional trust | 18:27 |
| rm_work | I actually just fixed that | 18:28 |
| rm_work | sec | 18:28 |
| blogan | well still possible to do but less straight forward | 18:28 |
| rm_work | yeah the Trust needs to be stored in the DB on the *tenant* not on the LB | 18:28 |
| rm_work | http://goo.gl/zzEDqN | 18:28 |
| xgerman | makes sense | 18:28 |
| blogan | so we'd need anothert able then | 18:28 |
| rm_work | do we not already store any tenant info? | 18:29 |
| blogan | store it based on entities | 18:29 |
| rm_work | or is it literally just the tenant_id in the LB table? | 18:29 |
| blogan | yep | 18:29 |
| rm_work | hmm | 18:29 |
| blogan | neutron may have one | 18:29 |
| rm_work | yeah then we'd need an additional table | 18:29 |
| blogan | of course that will then bring up whether it should be a neutron wide table or lbaas only | 18:30 |
| rm_work | hmm | 18:30 |
| rm_work | well | 18:30 |
| rm_work | if we're planning to spin out, this is another place we'd want to not mix stuff with neutron | 18:30 |
| blogan | good point | 18:30 |
| rm_work | though if it's in common utils... | 18:30 |
| rm_work | which i think it is... blah | 18:30 |
| blogan | lol | 18:30 |
| blogan | yeah | 18:30 |
| rm_work | that sucks | 18:30 |
| xgerman | should be neutron wide ideally | 18:30 |
| blogan | when we spin out we'll need to just make sure we have everything we need | 18:30 |
| rm_work | we're putting all the barbican handling into neutron/common | 18:31 |
| rm_work | so we'll just need to clone it | 18:31 |
| rm_work | when we spin out :/ | 18:31 |
| blogan | pretty much | 18:31 |
| rm_work | sooo | 18:31 |
| rm_work | for now the trustID would need to be stored as part of neutron-core | 18:31 |
| rm_work | when we spin out we'll need to add a table for it | 18:31 |
| blogan | well octavia will most likely need a table for it anyway right? | 18:31 |
| rm_work | yes, octavia also | 18:31 |
| rm_work | err no | 18:31 |
| rm_work | Octavia won't use trusts | 18:32 |
| rm_work | so it does not | 18:32 |
| blogan | oh thats right | 18:32 |
| *** RaginBajin has quit IRC | 18:32 | |
| *** mestery has quit IRC | 18:32 | |
| rm_work | you confused me for a sec :P | 18:32 |
| blogan | storing it all under its service account | 18:32 |
| rm_work | yep | 18:32 |
| xgerman | that's something we might need to revisit | 18:32 |
| *** mestery has joined #openstack-lbaas | 18:32 | |
| rm_work | probably | 18:32 |
| blogan | im sure when the time comes to implement, it'll be a good discussion | 18:33 |
| xgerman | yep | 18:33 |
| rm_work | but for now, this is workable, yes? | 18:33 |
| blogan | yeah | 18:33 |
| xgerman | +1 | 18:33 |
| rm_work | though, I am about to have a discussion with some keystone folks, once their meeting is over | 18:33 |
| blogan | and then you'll find out its not workable | 18:33 |
| rm_work | heh | 18:33 |
| rm_work | well they said it "works" | 18:33 |
| xgerman | just assume it is :-) | 18:33 |
| blogan | did they use air quotes? | 18:33 |
| rm_work | but they are concerned about the creation of the Trust | 18:33 |
| rm_work | they would rather the user create the trust before they deal with us | 18:34 |
| rm_work | which I agree is better from a security perspective | 18:34 |
| rm_work | but often security loses to ease of use :P | 18:34 |
| blogan | ahh i see, that is definitely soemthing to think about | 18:34 |
| xgerman | they just need to learn from Facebook - this app Lb canb access your credit card, etc. | 18:34 |
| xgerman | Ok? | 18:34 |
| rm_work | yeah, I was thinking about exactly that from the beginning, but opted to go with this model so the user wouldn't need to figure out all the Trust bs | 18:35 |
| blogan | doing something on the behalf of the user is always somethign to steer away if not needed | 18:35 |
| rm_work | yes, we'll see | 18:35 |
| blogan | it would be odd to tell a user that htey first have to tell keystone to set up a trust with their account to this service lbaas account | 18:35 |
| blogan | if it was an admin doing it, it would be fine, but a common user? | 18:36 |
| xgerman | yep, that's just the opposite what Facebook + Google do | 18:36 |
| blogan | and they do no evil | 18:36 |
| blogan | so we'd be on the side of not evil | 18:36 |
| xgerman | exactly -- | 18:37 |
| blogan | but not exactly good | 18:37 |
| xgerman | nah, they can always revoke the trust | 18:38 |
| *** RaginBajin has joined #openstack-lbaas | 18:55 | |
| *** sbalukoff has joined #openstack-lbaas | 18:57 | |
| *** openstackgerrit has quit IRC | 19:02 | |
| *** jroyall has joined #openstack-lbaas | 19:13 | |
| *** jroyall is now known as joeroyall | 19:13 | |
| *** mlavalle has quit IRC | 19:17 | |
| *** mlavalle has joined #openstack-lbaas | 19:18 | |
| *** vivek-ebay has quit IRC | 19:18 | |
| *** vivek-ebay has joined #openstack-lbaas | 19:19 | |
| *** ptoohill-oo has joined #openstack-lbaas | 19:33 | |
| *** ptoohill has quit IRC | 19:34 | |
| *** openstackgerrit has joined #openstack-lbaas | 19:41 | |
| *** vivek-ebay has quit IRC | 19:43 | |
| *** joeroyall has quit IRC | 19:57 | |
| *** mikedillion has quit IRC | 20:25 | |
| *** sbalukoff1 has joined #openstack-lbaas | 20:34 | |
| *** sbalukoff has quit IRC | 20:35 | |
| *** sbalukoff1 is now known as sbalukoff | 20:38 | |
| *** TrevorV_ has quit IRC | 20:42 | |
| *** RaginBajin has quit IRC | 20:45 | |
| rm_work | dougwig / sbalukoff / xgerman / blogan: so, how do we feel about giving the user a template (like, a pre-filled JSON body) and requiring them to hit Keystone and set up the Trust, before they create the TLS Termination? | 20:55 |
| rm_work | instead of setting up the trust ourselves by hijacking their token during the TLS Termination Create operation? | 20:56 |
| dougwig | i feel pretty yuck about that. | 20:56 |
| rm_work | Keystone people feel very strongly about it | 20:56 |
| rm_work | I tend to agree with you though, dougwig | 20:56 |
| rm_work | so it may come down to us ignoring what people on Keystone tell us :/ | 20:56 |
| dougwig | use AWS or backspace as examples of what people expect from load balancer setup. that's not anywhere in the mix. | 20:56 |
| *** mestery has quit IRC | 20:57 | |
| rm_work | lol, backspace | 20:57 |
| dougwig | well, it's a correct "keystone is the center of the universe" approach. it just makes no sense for end users. | 20:57 |
| *** mestery has joined #openstack-lbaas | 20:57 | |
| dougwig | damn this dumbed down autocorrect. :) | 20:57 |
| rm_work | of course, neither is sharing a cert/key from an external repo, so | 20:57 |
| rm_work | this is all new :P | 20:57 |
| rm_work | yeah... so, one thing I am interested in, is whether "having a thing in Barbican" is going to cost the user money | 20:58 |
| rm_work | if that's billable, then we're adding that cost to them | 20:58 |
| dougwig | well, you trust certs from your registrar, without setting up trusts. that's got an easier paradigm/flow. plus, i'll assume that horizon will make it look semi-seamless, like they do with ssh keys and nova. | 20:58 |
| rm_work | but that's just a random side-thought, not exactly relevant now | 20:58 |
| dougwig | rm_work: good point, though i'd pay money to get auto-renewals on expiration. | 20:59 |
| rm_work | yeah, probably | 20:59 |
| rm_work | dougwig: yes, me too! but we're not quite there yet :) | 20:59 |
| rm_work | that requires a few more pieces to fall into place | 20:59 |
| rm_work | but is definitely the end goal I think | 20:59 |
| rm_work | since Barbican can now get Certs generated by Symantec | 21:00 |
| sbalukoff | I think what the Keystone people want is too jarring / confusing to the user. | 21:00 |
| rm_work | sbalukoff: agree | 21:00 |
| *** jorgem has quit IRC | 21:00 | |
| rm_work | but the one complication about "just ignoring them" is they said something about the user's key not always allowing Trust Creation | 21:00 |
| rm_work | so if that is common, it's a problem for us | 21:00 |
| sbalukoff | I imagine we'd end up a with a lot of complaints from people who can't figure out how to add their damned TLS key. :P | 21:01 |
| rm_work | sbalukoff: also agree :P | 21:01 |
| sbalukoff | rm_work: If we encounter that, we return an error. | 21:01 |
| dougwig | i sometimes get the feeling that a lot of folks in openstack are assuming more of the world uses orchestration than is reality. | 21:01 |
| sbalukoff | dougwig: Indeed! | 21:02 |
| dougwig | because so many of these integration points are being punted away with heavy-weight alternatives that pretty much *have* to be scripted to make any sense. | 21:02 |
| sbalukoff | Also folks: Right now there's not a whole lot on the agenda for tomorrow, though I think we could probably spend most of the meeting just debating using Storyboard for better blueprint / project management. | 21:02 |
| sbalukoff | Nevertheless, if y'all have things you want to add, feel free to add them to the agenda. | 21:02 |
| sbalukoff | (I don't think it's worthwhile to send a message reminding people about tomorrow's meeting to the mailing list-- I've done that for several weeks now and it hasn't really changed attendance. So if they haven't figured it out by now...) | 21:03 |
| dougwig | summon the IRC supporting flock! | 21:04 |
| sbalukoff | dougwig: Er... ? | 21:04 |
| dougwig | the vote tomorrow. | 21:04 |
| sbalukoff | dougwig: Actually, I've been recruiting heavily among people at Blue Box to attend / start contributing to Octavia. So far, it seems most if not all I've convinced to come are heavily in favor of keeping the meetings in IRC. | 21:05 |
| dougwig | of course they are; it's the clearly superior medium. | 21:05 |
| sbalukoff | I'm also somewhat more amenable to the idea-- to the point where I might just abstain from voting. | 21:05 |
| dougwig | ;-) | 21:05 |
| sbalukoff | dougwig: Are you trying to get me to vote against IRC? | 21:05 |
| dougwig | would i ever troll? | 21:05 |
| sbalukoff | Well, you never use sarcasm. | 21:06 |
| sbalukoff | And I'd rather die than use hyperbole. | 21:06 |
| dougwig | both of which are excellent uses of a text-only medium, too. | 21:06 |
| sbalukoff | Haha | 21:06 |
| sbalukoff | As a friendly reminder, here's the agenda so far: https://wiki.openstack.org/wiki/Octavia/Weekly_Meeting_Agenda#Agenda | 21:07 |
| blogan | douwig: backspace == racklink | 21:16 |
| blogan | sbalukoff: wait you were convinced to use IRC by your own people? but we couldn't convince you? | 21:17 |
| sbalukoff | blogan: I wasn't convinced to use IRC per se. | 21:17 |
| sbalukoff | But I'm more on the fence about it. | 21:17 |
| sbalukoff | Now, anyway. | 21:17 |
| blogan | i'd like to meet these people that pushed you to be on the fence, do they ahve some kind of mind control? | 21:18 |
| sbalukoff | I forgot my tinfoil hat that day. | 21:18 |
| sbalukoff | At least one of them will probably be joining us tomorrow. | 21:18 |
| blogan | i've got a few questions, api related, but some don't have any real solutions other than adding features to another library | 21:19 |
| sbalukoff | For the meeting, you mean? | 21:19 |
| sbalukoff | Add an agenda item! | 21:19 |
| blogan | yaeh | 21:20 |
| sbalukoff | Or are you talking about discussing them here and now in IRC? | 21:20 |
| blogan | doesn't matter, though I think the more people the more likely they ahve experience | 21:20 |
| blogan | you have any experience with WSME? | 21:20 |
| sbalukoff | Nope. | 21:20 |
| blogan | so i guess the one question is that it looks like openstack is drfiting towards a pecan + wsme standard for APIs | 21:21 |
| blogan | just wanted to make sure everyone is on board with that | 21:21 |
| blogan | also that we should use jsonschema | 21:21 |
| sbalukoff | Cool. Add the agenda item. | 21:21 |
| blogan | i have to type? | 21:22 |
| blogan | yuck | 21:22 |
| sbalukoff | ... | 21:22 |
| sbalukoff | Just because you're in Texas doesn't mean I don't have ways to smack you. | 21:22 |
| dougwig | i'm on board with pecan/wsme. i hate json schemas; i think they imply a lazy interface design, and they're ugly. but i'd live. | 21:22 |
| blogan | dougwig: I like it and I don't | 21:23 |
| dougwig | that's a good summary | 21:23 |
| blogan | there is some overlap with with WSME can do as well | 21:23 |
| blogan | lol | 21:23 |
| blogan | so if people don't want to use jsonschema, we can use WSME to do some of the things | 21:24 |
| blogan | though I'm not sure WSME will validate values | 21:25 |
| blogan | still learning about it | 21:25 |
| rm_work | well, if you're talking about validation of JSON, take a look at how Barbican does it: https://github.com/openstack/barbican/blob/master/barbican/common/validators.py | 21:26 |
| blogan | ive seen it | 21:26 |
| rm_work | which method do they use? | 21:26 |
| rm_work | I don't actually know what jsonschema is | 21:27 |
| blogan | jsonschema | 21:27 |
| rm_work | is that a python lib? | 21:27 |
| rm_work | is that what they are using? | 21:27 |
| blogan | and also some customs | 21:27 |
| rm_work | ah, yeah it is, I see the import now T_T | 21:27 |
| dougwig | it's a draft RFC for encoding a schema inside json. | 21:27 |
| rm_work | well, I didn't mind their validator code -- writing it seemed straightforward | 21:27 |
| rm_work | and i believe it works fine | 21:27 |
| rm_work | if I had to vote between that and doing it by hand, I might vote for jsonschema | 21:28 |
| blogan | i dont mind using it, or using whatever wsme does as well | 21:28 |
| dougwig | rm_work: until you saw a jsonschema, maybe. | 21:28 |
| blogan | they're in that link | 21:28 |
| blogan | it makes sense to me | 21:28 |
| blogan | actually very similar to the extension dict in neutron | 21:29 |
| blogan | except, it thankfully doesn't also define the resources and methods the plugin should define | 21:29 |
| blogan | dougwig: is you're preference to write our own validation? | 21:29 |
| blogan | or just use whatever wsme does and add in what it is lacking | 21:30 |
| blogan | WSME is moving to stackforge so we can definitely make that much better | 21:30 |
| dougwig | my *preference* is to use a framework like django or rails to automate all this cruft away. in an openstack context, i'm happy to settle for doing what is the norm within this community. | 21:31 |
| rm_work | dougwig: yeah, I apparently wrote jsonschema validators without even realizing it, and i liked them | 21:31 |
| blogan | i know pecan and wsme are becoming the norm, but jsonschema maybe not, though it is being used | 21:32 |
| blogan | obviously barbican is one | 21:32 |
| blogan | that is | 21:32 |
| blogan | but they're not using WSME | 21:32 |
| dougwig | it'd be better than no validation or hand validation, for sure. | 21:32 |
| blogan | i think improving WSME's validation would be the best way | 21:33 |
| blogan | but, a stop gap would be to use jsonschema | 21:33 |
| xgerman | I saw barbican using flask.. | 21:37 |
| xgerman | or was it uwsgi | 21:37 |
| xgerman | always confuse things | 21:37 |
| *** vivek-ebay has joined #openstack-lbaas | 21:38 | |
| xgerman | but we know rmworks opinion to uwsgi :-) | 21:38 |
| rm_work | yeah, uwsgi <_< | 21:38 |
| rm_work | well, it has problems | 21:39 |
| rm_work | but | 21:39 |
| rm_work | for *dev only depolyments* it's not a bad option I guess | 21:39 |
| dougwig | that's only if it's the edge server. you can put a reverse proxy in front of it and it works fine. | 21:40 |
| *** vivek-ebay has quit IRC | 21:42 | |
| rm_work | yeah, the uwsgi dev flat out told me it's not intended to ever be exposed directly | 21:43 |
| xgerman | I like uwsgi + flask -- but we should likely follow their standard | 21:46 |
| rm_work | I like Flask a whole lot better than Pecan, but Pecan is the Openstack framework of choice, so I think that's what we're stuck with, for better or for worse | 21:47 |
| *** vivek-ebay has joined #openstack-lbaas | 21:50 | |
| blogan | xgerman: doesn't libra use pecan? | 21:52 |
| xgerman | well, we use pecan and wsme or something on top | 21:53 |
| blogan | well right now, for simplicity, im just using wsgiref.simple_server as the server | 21:53 |
| blogan | but it can be swapped out easily | 21:54 |
| xgerman | but I like flask better :-) | 21:54 |
| blogan | well you cant have it! | 21:54 |
| xgerman | wsme also runs on top of flask | 21:54 |
| blogan | wsme runs on top of all the frameworks from what i can tell | 21:54 |
| blogan | lets use falcon | 21:55 |
| xgerman | yeah, wsme seems to bring out the wrost from each framework :-) | 21:57 |
| xgerman | also we can likely diverge for the REST server on the amp | 21:58 |
| blogan | oh yeah definitely | 21:59 |
| xgerman | just use them all in some way or another | 21:59 |
| dougwig | just make it ruby and call it a day. | 22:00 |
| xgerman | Remember Rackspace tried Java | 22:01 |
| xgerman | but I think you can always use jython | 22:01 |
| blogan | vba | 22:01 |
| blogan | microsoft office load balancer | 22:02 |
| xgerman | that's discontinued. Iron Ruby | 22:02 |
| sbalukoff | We just went with a simple CGI script as the agent on our devices. Didn't need anything more than that because the back-end API was so simple. | 22:02 |
| sbalukoff | (lighttpd as the webserver) | 22:02 |
| rm_work | ooh, let's do that with a php script | 22:02 |
| sbalukoff | Ick. | 22:02 |
| dougwig | SOAP with xmlsig. | 22:02 |
| rm_work | haha | 22:02 |
| dougwig | that's a thread winner right there. | 22:03 |
| sbalukoff | Ugh. | 22:03 |
| xgerman | let's sue one of those new languages, golang, hack, ... | 22:03 |
| blogan | do we have a weekly etherpad to update? | 22:04 |
| xgerman | we do | 22:04 |
| xgerman | sbalukoff?? | 22:04 |
| sbalukoff | blogan: Yes, we do | 22:05 |
| sbalukoff | https://etherpad.openstack.org/p/octavia-weekly-standup | 22:05 |
| blogan | i still have the etherpad, but i'll let you add the template | 22:05 |
| sbalukoff | Er... there was a template on the old one? | 22:05 |
| blogan | well new week section | 22:05 |
| sbalukoff | Aah. Ok. | 22:06 |
| blogan | not exactly a template | 22:06 |
| sbalukoff | Sure. | 22:06 |
| *** vivek-ebay has quit IRC | 22:14 | |
| rm_work | hmm, how do I add things to the *Neutron-LBaaS* meeting agenda? | 22:17 |
| rm_work | do I just talk to dougwig or jorgem? | 22:17 |
| rm_work | or is there a wiki | 22:17 |
| blogan | there is a wiki, but it hasnt been updated in a while | 22:18 |
| blogan | well 2 or 3 weeks | 22:18 |
| rm_work | yeah, looked that way | 22:18 |
| dougwig | just edit the wiki | 22:18 |
| rm_work | not sure how the last few agendas got set | 22:18 |
| blogan | wasn't much to talk about so it was mostly open discussion | 22:18 |
| rm_work | ok, mayhaps I will just do that <_< | 22:18 |
| blogan | basically there's the obligator "incubator update" then something I probably added just for the hell of it, and then open discussion | 22:19 |
| rm_work | k | 22:19 |
| dougwig | there's not much to discuss while we're in limbo, and octavia has the focus. | 22:19 |
| xgerman | we might want to talk design summit again | 22:19 |
| xgerman | and jorgem's election campaign to become PTL of Neutron - kidding... | 22:20 |
| blogan | when are those going to be open for registering? | 22:20 |
| blogan | he woudl lose in a landslide | 22:20 |
| xgerman | that's before the campaign | 22:20 |
| blogan | he does play dirty | 22:20 |
| sbalukoff | Question for y'all: With the discussion of pecan / WSME, would you rather see the haproxy amphora agent API proposal I have in mind as stubs in code, or as .rst documentation that will be incorporated into Octavia's official docs? | 22:20 |
| sbalukoff | Waht I mean by that... | 22:20 |
| dougwig | are you trying to get out of actual python coding? | 22:21 |
| sbalukoff | The former is great if there's an automatic way for documentation to be generated. But I think it's good to have the latter in any case. | 22:21 |
| dougwig | :) | 22:21 |
| blogan | lol | 22:21 |
| sbalukoff | dougwig: If I do the former, it will take me 2-3 x as long, probably. Because, yes, I'm a terrible python coder. :) | 22:21 |
| blogan | sbalukoff: is your intention to become a better python coder? | 22:21 |
| dougwig | either is fine. i'm sure one of us can transcribe to python quickly. | 22:22 |
| sbalukoff | blogan: Honestly, I'm not sure: I think there are enough contributors to this project that are light-years ahead of me as far as their python coding ability. Even if I stopped doing anything else for the next three months and just coded python all day I wouldn't catch up to those of y'all who have years of experience in this... | 22:23 |
| sbalukoff | But I think this (and other Open Source projects) often lack good... I dunno... project management? | 22:23 |
| sbalukoff | I feel like the biggest challenge Octavia has right now isn't my ability to write shitty code, it's that people have a real hard time knowing where to start eating this elephant. | 22:24 |
| sbalukoff | If that makes sense. | 22:24 |
| blogan | eating elephants makes all the sense to me | 22:24 |
| sbalukoff | I hesistate to call that project management because I think that term carries a lot of baggage. | 22:24 |
| xgerman | I think we need some more architecture/design discipline | 22:25 |
| blogan | its up to you on what you want to do any coding, though I think being the PTL it would help you out in knowing the code | 22:26 |
| sbalukoff | Haha | 22:26 |
| sbalukoff | Of course-- I can read it just fine. It's not too dissimilar from perl or ruby. | 22:26 |
| xgerman | yeah, what I am saying :-) | 22:26 |
| sbalukoff | And I try to be thorough on code reviews. | 22:27 |
| blogan | still I think its beneficial to know how to do it, you could start on it, even if it'd take you 2-3x times longer once it is needed someone could come in and give it a boost | 22:28 |
| sbalukoff | Anyway, maybe it's more a question of: Am I blocking anyone else's work if I take a while to come up with the haproxy amphora agent API, doing it in code (which, again, I'm hoping will be something from which automatic documentation can be generated, as much as I dislike that term.) | 22:28 |
| dougwig | if you do this one as a doc, because abstract interfaces aren't the best way to learn, you ought to make up for it by writing some unit tests or something. or else we will all mock you. yeah, check out that wicked pun. | 22:28 |
| blogan | lol | 22:28 |
| sbalukoff | Haha | 22:28 |
| sbalukoff | Right. | 22:29 |
| blogan | sbalukoff = mock.Mock() | 22:29 |
| sbalukoff | But... not until after I convince you all to go with Storyboard because launchpad sucks. | 22:29 |
| sbalukoff | ;) | 22:29 |
| blogan | sbalukoff.assertCalledWithAnger() | 22:29 |
| dougwig | abstract interfaces (import abc) + docstrings == a doc, btw. | 22:29 |
| sbalukoff | (though I'm serious about moving to storyboard. It's not the best tool I've seen, but it's way, way better than launchpad.) | 22:29 |
| xgerman | johnsom says he would have a blueprint by the end of the week for the controller. Biy this wpould eb driver <-> controller? | 22:29 |
| dougwig | we can replace assertTrue with assertRighteous | 22:29 |
| xgerman | johnsom_ | 22:30 |
| blogan | i think the interface is much cleaner, but i'm not sure it really adds any other value | 22:30 |
| johnsom_ | I think we need to have .rst specs in the blueprints. It will help our OpenStack aspirations | 22:30 |
| sbalukoff | dougwig: Right, I just didn't know whether there was an automated way to extract those docstrings and have them populated in the "standard" build documentation. | 22:30 |
| xgerman | +1 | 22:30 |
| sbalukoff | I guess I could figure that out. | 22:30 |
| dougwig | sbalukoff: indeed there is, though i've never done it. but the python interface docs are built that way. | 22:30 |
| johnsom_ | I said I would have a start on the controller blueprint this week. | 22:30 |
| blogan | the problem with having specs right now is that we'd need to draft them up, then review the specs, then implement the code, then review the code | 22:31 |
| xgerman | that's how software dev works | 22:31 |
| sbalukoff | blogan: +1 | 22:31 |
| johnsom_ | Yes, which is a demonstration of the openstack process and we have our act together... | 22:31 |
| blogan | yes the openstack process on projects that are mature | 22:32 |
| sbalukoff | And it makes things move approximatly 1/10th the speed while we are still bootstrapping. | 22:32 |
| blogan | which i agree we should do once this is in a mature enough state | 22:32 |
| sbalukoff | Yep. | 22:32 |
| xgerman | well, we want to be an OpenStack project so we need to demonstrate their pocesses. What's the alternative? Drinking a bottle of tequial and writing LBaaS? | 22:33 |
| johnsom_ | I don't see how we can get to a mature state without a good set of reviewed specifications | 22:33 |
| xgerman | +1 | 22:33 |
| dougwig | as a group, we spend a *lot* of cycles worrying about being in openstack. seemingly more than we spend building something that we can put in openstack. is the current bootstrapping leaving anyone feeling like they don't know what's going on, or that we're not iterating in the right direction? | 22:33 |
| sbalukoff | xgerman: I see what you did there! | 22:33 |
| sbalukoff | dougwig: +1 | 22:34 |
| sbalukoff | I feel like we're at the "write some goddamn code already" stage in many ways. | 22:34 |
| johnsom_ | dougwig: yes, I don't feel like we have a solid direction | 22:34 |
| sbalukoff | dougwig: Yes, that's my impression as well. | 22:34 |
| sbalukoff | I think this could be helped by fleshing out the current blueprints. | 22:35 |
| *** jroyall has joined #openstack-lbaas | 22:35 | |
| *** jroyall is now known as joeroyall | 22:35 | |
| sbalukoff | But I'd like to move to a better system before we do that (ie. Storyboard) | 22:35 |
| johnsom_ | Why do blueprints at all if we aren't going to build specs behind them and review? | 22:35 |
| sbalukoff | Also, I'm starting to be convinced that storyboard is the "future" of how blueprinting is going to work for OpenStack anyway... so I'd rather get on it now rather than lose a bunch of history in launchpad by doing a transition later. | 22:36 |
| dougwig | johnsom_: blueprints are a way for the manager to make a schedule/milestones. specs are to get devs talking before coding. i'm feeling like things are early enough that the code reviews are simple enough to be the rallying point for discussion. i take it that you disagree? | 22:37 |
| sbalukoff | johnsom_: I think blueprints at this stage are a good way to show logical divisions in work domain-- so that people can work independently in many cases without worrying about stepping on each others toes. | 22:37 |
| blogan | plus we agreed that when you pick up a blueprint in launchpad, you'd add more details to it explaining at a high level of how to accomplish the implementation | 22:38 |
| sbalukoff | blogan: Except I don't think people know what details to fill out. | 22:38 |
| sbalukoff | blogan: I think we're going to have to do that for people so they know what the blueprint actually entails. | 22:38 |
| blogan | no they don't, but do people know what details to fill out in a blueprint spec? | 22:38 |
| xgerman | yep, we should establish good processes/docs | 22:39 |
| sbalukoff | We're eating an elephant, but even knowing how to eat an elephant leg seems overwhelming. | 22:39 |
| johnsom_ | blogan: yes, I think that is what I am arguing for, is filling out the specs behind those blueprints. It will help with the integration points in the code. | 22:39 |
| sbalukoff | And again, the launchpad system really doesn't help. :P | 22:39 |
| blogan | no it doesn't, but even the blueprint spec process is pretty new to openstack anyway and so most projects (if any) were not boostrapped by doing specs | 22:39 |
| sbalukoff | that's true. | 22:40 |
| blogan | specs being similar to the neutron-specs | 22:40 |
| sbalukoff | They were mostly bootstrapped by a small group of devs working together who shared a common vision and just sat down and wrote code. | 22:40 |
| blogan | now giving as many details int he description/work items in launchpad section is great, and bring it up in the weekly meetings | 22:40 |
| sbalukoff | I feel like with the v0.5 discussion, hopefully people have enough of that common vision. | 22:40 |
| johnsom_ | I have spent some time looking at the blueprints for "mature" projects to get a feel for how they are used. I am coming up to speed on them and plan to post some specs in the next week or so | 22:40 |
| dougwig | is the issue here that we need specs, or that we need to get the direction of each component less abstract/more defined before some are comfortable running with it? (and a spec is one way of accomplishing that.) | 22:41 |
| blogan | johnsom_: specs on blueprints you are doing? | 22:41 |
| sbalukoff | But I'm discovering with people I'm recruiting to work on this internally, that they still need more specifics than "write controller" to get started even knowing how to break that up into smaller tasks. | 22:41 |
| blogan | sbalukoff: oh thats definitely a problem, and really I didn't know exactly what that woudl entail with the controller either, though I knew the controller has the most dependencies | 22:42 |
| johnsom_ | blogan: I signed up for base-image and controller | 22:42 |
| sbalukoff | dougwig: I think the problem is that each of these blueprints need better definitions, probably from me or blogan (or others here who have a good sense of how things fit together.) | 22:42 |
| sbalukoff | johnsom_: I suspect that's too much work for one person. | 22:42 |
| dougwig | johnsom_: heh, both?? :) | 22:42 |
| johnsom_ | base-image is pretty straight forward | 22:43 |
| sbalukoff | Considering that's probably *most* of the code that will be written. ;) | 22:43 |
| blogan | sbalukoff: that was the point of me dong them right now, to get it started and then as things became more clear (or unclear) then more details would be added, blueprints would be split into multiple ones | 22:43 |
| sbalukoff | blogan: Sure, but I think we should probably fill in the bits we can see already. This will help those who don't see those bits yet. | 22:43 |
| blogan | perfectly fine by me | 22:44 |
| sbalukoff | (while being open to additions or deletions from those bits) | 22:44 |
| johnsom_ | sbalukoff: we are spinning up a team, so I will have help | 22:44 |
| sbalukoff | Ok, again, this could be done using launchpad, but I think storyboard is much better suited to the task. | 22:44 |
| sbalukoff | johnsom_: Then I'd rather have visibility into that team. Get your team-mates to join here and in the meetings, eh! | 22:44 |
| blogan | sbalukoff: im just concerned with the adoption of storyboard, doesn't seem like its being adopted at a high rate | 22:45 |
| dougwig | pushing storyboard before even infra is using it for real seems.... like some serious dislike for launchpad. | 22:45 |
| xgerman | also to add that johnsom_ is a god | 22:45 |
| blogan | and i don't see any extra added value right now other than it being a cleaner UI | 22:45 |
| sbalukoff | blogan: Full disclosure, I've got the guy who works on refstack (David Lenwell) helping me out on this. :D | 22:46 |
| xgerman | also we are a big team so we need process | 22:46 |
| xgerman | (we = Octavia people) | 22:46 |
| johnsom_ | sbalukoff: you will see that Al has already signed up for some work. | 22:46 |
| sbalukoff | Also, storyboard is run by the openstack-infra guys. | 22:46 |
| johnsom_ | xgerman: you are funny | 22:46 |
| sbalukoff | And is much more responsive for bugfixes / feature additions than launchpad. :D | 22:46 |
| sbalukoff | blogan: You can split blueprints into tasks. | 22:47 |
| sbalukoff | And you can comment on blueprints | 22:47 |
| blogan | what if it is never adopted because openstack has a knack for adopting new technologies? | 22:47 |
| xgerman | well, let's not argue tools | 22:47 |
| dougwig | well, sbalukoff is going to have to use those tools for organizing more than the rest of us, so i'm fine with him picking his poison. | 22:47 |
| blogan | slowly | 22:47 |
| blogan | i really don't care iether way, just voicing concerns | 22:48 |
| johnsom_ | It seems like we have a start in launchpad, maybe we should let storyboard cook a bit and switch on the next round | 22:49 |
| xgerman | my main concern is that we need some process to get design documents -- we are far too many people to just code and it will work out | 22:49 |
| sbalukoff | Got it. I can't predict the future, but it certainly seems to me that launchpad is obtuse enough, and poorly designed enough, that something like storyboard is likely to supersede it. | 22:49 |
| blogan | adding comments to blueprints is actually a huge feature, if anyone can add comments | 22:49 |
| blogan | that way we can actually get feedback on the blueprint iteself | 22:49 |
| sbalukoff | blogan: That's the first thing I latched onto. | 22:50 |
| sbalukoff | Also, when you split a blueprint up into tasks, you can assign those tasks to different people easily. | 22:50 |
| sbalukoff | So, it helps foster more people contributing in parallel. | 22:50 |
| sbalukoff | Each of those tasks has a status... | 22:50 |
| sbalukoff | So you can see, pretty quickly, where things are at right now with the blueprint. | 22:51 |
| sbalukoff | (Also, a pretty extensive history) | 22:51 |
| blogan | xgerman: i agree with you there, but I still think too much process in the bootstrapping phase will slow this down to a snail's pace, look how long it took us to agree on the name of the device hosting the load balancer | 22:51 |
| sbalukoff | I don't think launchpad provides any of this. And if it does, then I predict its buried somewhere difficult to find. | 22:51 |
| sbalukoff | xgerman: I don't have a problem with people writing specs especially around APIs and interfaces between components... | 22:52 |
| sbalukoff | I think that's a really good idea. | 22:52 |
| rm_you | oh man, you guys generated like.... half of my visible scrollback, in the 30 minutes i was driving home >_< | 22:52 |
| sbalukoff | But for other things, sometimes just having some code to discuss can move us forward fastr. | 22:52 |
| xgerman | lol | 22:52 |
| sbalukoff | You know me: I'm a huge fan of documentation. :) | 22:52 |
| sbalukoff | rm_you: Then our nefarious plan worked! | 22:52 |
| xgerman | well, we have some fundamental decisions to make which will hunt us for many years -- so slower speed and getting it right is better for me right now | 22:53 |
| sbalukoff | xgerman: I might point out that that's a bit of a false dichotomy. :/ | 22:53 |
| johnsom_ | I just see broken pages (dashboard), pages that spin loading, and the lack of subscribe (unless I missed it) features.... | 22:54 |
| blogan | xgerman: i don't think it will haunt us for many years because we can easily change it since it is just bootstrapping | 22:54 |
| dougwig | we have warring software methodologies going on here. spec like mad up front and hope you get it right before coding, or write some code, knowing some will be crap, and iterate. | 22:54 |
| sbalukoff | johnsom_: What browser are you using? It's all working fine for me (using Chrome on a Mac) | 22:54 |
| dougwig | aka waterfall vs agile | 22:54 |
| johnsom_ | Chrome | 22:55 |
| sbalukoff | dougwig: I think we need a happy-medium between both. | 22:55 |
| sbalukoff | If that makes sense. | 22:55 |
| blogan | agilefall | 22:55 |
| xgerman | I am ok with just coding the diretcoy structure, coding the controller not so much | 22:55 |
| rm_you | It looked fine to me (also Chrome/OSX) but I wasn't quite sure what the point was | 22:56 |
| rm_you | I guess I need to see Storyboard in action | 22:56 |
| sbalukoff | dougwig: That is to say, if you're not sure how to do something, discuss it, spec it up, etc. But if you know what you want to do, write the code and see what people think. | 22:56 |
| rm_you | (for a project I understand somewhat) | 22:56 |
| sbalukoff | rm_you: Poke around in the refstack project. | 22:56 |
| johnsom_ | I definitely see the benefit to the history and easier task assignment. I'm just not sure how cooked it is and wish it had the subscribe feature. | 22:57 |
| sbalukoff | FWIW, storyboard encourages using specs. :) | 22:57 |
| xgerman | sbalukoff, I have a clear idea that I don't want a driver in the controller so I sahould just code it that way? | 22:57 |
| blogan | xgerman: thats been discussed | 22:57 |
| johnsom_ | It is also missing the link to specs | 22:57 |
| blogan | xgerman: so not an accurate counter | 22:57 |
| sbalukoff | xgerman: Right, and then we'd shoot you down in the code review. ;) Especially because we have been discussing this and will be voting on it. | 22:57 |
| sbalukoff | xgerman: So yes, those kinds of decisions are worth having a discussion about before code. | 22:58 |
| xgerman | blogan, I just discussed it because I didn't know I could just write it and didn't need consensus... | 22:58 |
| sbalukoff | xgerman: But if you are better at expressiong your idea through code, I think that's valid. | 22:58 |
| sbalukoff | xgerman: That's where the 'iteration' bit comes in. | 22:58 |
| johnsom_ | sbalukoff: where is the spec tie-in with storyboard, I am missing it | 22:58 |
| xgerman | I am just saying we need some spec before we write a ton of code we will -2 | 22:59 |
| sbalukoff | johnsom_: Oh, looks like it's usually added just as a first task. So yes, I guess that's up to whoever is running the project. (example: https://storyboard.openstack.org/#!/story/105 ) | 23:00 |
| dougwig | xgerman: i'm not sure any of the reviews should be a "ton" of code yet... maybe that's our disconnect? | 23:00 |
| blogan | sbalukoff: lol so it really doesn't encourage specs, its whomever is creating the tasks | 23:00 |
| sbalukoff | blogan: Hmmm... David seems to imply that it did, though, like you, I'm not seeing obviously how it does. | 23:01 |
| dougwig | i don't have a problem with specs. i only don't like them when writing the spec takes as long as writing the code, which is usually true of smaller bites. nor do i mind a -2 in such a situation. if you're writing a big thing and not sending any of it to gerrit until the end, then a spec would be a damn fine idea; but i'd also question that approach. | 23:01 |
| johnsom_ | sbalukoff: hmm, I was looking on the openstack-infra/storyboard project and it doesn't have that. In fact, task 1 is missing altogether. | 23:02 |
| xgerman | dougwig, agreed. But I also see value in parcticing to write specs with small parts | 23:02 |
| xgerman | especially since we are settling on new tools and most of us are new to them | 23:02 |
| xgerman | But I am ok with a spec just being the abstract class IHMO | 23:03 |
| sbalukoff | johnsom_: Right. So I retract my earlier statement: I'm also somewhat unfamiliar with this tool (though, again, I already like it a lot more than launchpad), and I don't see exactly how it encourages the writing of specs. | 23:03 |
| *** sbfox has quit IRC | 23:04 | |
| dougwig | xgerman: i don't have a problem with that, as long as we're being fluid about it and not just adding a required heavy-weight process just because. there are things that i'd want to spec out (like the entire controller) just to get my head straight and get some feedback quicker, and there are other things (like the neutron interface), where it'll just be | 23:04 |
| dougwig | easier to throw some code into gerrit and start getting comments there. | 23:04 |
| dougwig | xgerman: just saw the abstract class comment; right, agreed. | 23:04 |
| xgerman | ok, glad. sbalukoff seomtimes makes things sound very one sided so I had to object :-) | 23:05 |
| sbalukoff | xgerman: Objecting with me for the sake of objecting with me gets us nowhere, eh. :P | 23:06 |
| sbalukoff | xgerman: If you're being serious, eh. | 23:06 |
| dougwig | xgerman: i'd actually prefer we use the specs directory in our repo for the occasional spec. the review process is more formal. | 23:06 |
| sbalukoff | That's what the specs directory is for! | 23:07 |
| xgerman | great. I like to see more specs unless for very simple things -- not just the things we "think" we know how it's done | 23:07 |
| dougwig | has this just been 1000 lines of irc violent agreement? i feel like this started with a nugget of discontent from johnsom_, and we didn't actually change anything. | 23:07 |
| xgerman | it's is human nature to overestimate | 23:07 |
| sbalukoff | dougwig: I used this discontent to push my super-secret agenda to get us all using storyboard instead of launchpad. | 23:08 |
| xgerman | dougwig we do a lot of that. Maybe that's why we shoul ditrch irc an d have a permanent google hangout | 23:08 |
| sbalukoff | (In reality, it was appropriate to the topic.) | 23:08 |
| sbalukoff | (Since the launchpad blueprint system is pretty horrible.) | 23:09 |
| blogan | while we're at violent agreements, Amphora! | 23:09 |
| sbalukoff | xgerman: HAHA! Oh, man, let's get the holy wars started, eh! | 23:09 |
| johnsom_ | Wow | 23:09 |
| * dougwig cries. | 23:10 | |
| blogan | alright I'm out, ill be on later I'm sure | 23:10 |
| sbalukoff | blogan: Too late! I already documented that non-arbitrary arbitrary decision in the wiki, and as we all know, the wiki is immutable. | 23:10 |
| blogan | sbalukoff: it's immutable becasue I don't want to touch wiki markup | 23:10 |
| sbalukoff | blogan: Doesn't it have a WYSWIG edtor java app or something attached? | 23:11 |
| sbalukoff | Er... | 23:11 |
| sbalukoff | WYSIWYG | 23:11 |
| xgerman | I can added Wikis. Blogan just tell me what you need and I will hack it there with big fomts and all caps | 23:12 |
| sbalukoff | xgerman: Don't forget to make the text blink. | 23:13 |
| xgerman | yep | 23:13 |
| dougwig | and red. gotta be red. | 23:17 |
| sbalukoff | Ooh! Cycle the colors through all the colors of the rainbow. Rapidly. | 23:17 |
| sbalukoff | Does anyone here remember Butt Fugly Design? | 23:17 |
| xgerman | disco lights? | 23:18 |
| johnsom_ | Can we have it start playing Rick Astley music when the page loads? | 23:19 |
| sbalukoff | I used to love to gawk at that site: 100% W3C standards compliant HTML code, and completely unsuable with nausea-inducing design. | 23:19 |
| rm_you | I remember that this is the website of the professor at my college that *taught the web design course*: http://www.cs.trinity.edu/~thicks/ | 23:19 |
| rm_you | I REALLY hoped it was an attempt at humor | 23:19 |
| rm_you | but I am afraid it wasn't | 23:19 |
| sbalukoff | rm_you: HAHA! | 23:19 |
| rm_you | them gifs... | 23:19 |
| rm_you | oh man and the javascript dropdown navigation that used frames and could end up in a loop... | 23:20 |
| sbalukoff | Animated gifs are the future of the internet. | 23:20 |
| sbalukoff | Get with the 1990's, man! | 23:20 |
| sbalukoff | Still better than frontpage. | 23:21 |
| johnsom_ | rm_you: The "Dr. Web" is choice! | 23:21 |
| rm_you | uhhh | 23:21 |
| rm_you | so | 23:21 |
| rm_you | view-source:http://www.cs.trinity.edu/~thicks/Main-Pages/Home.html | 23:21 |
| rm_you | it IS frontpage | 23:21 |
| johnsom_ | <meta name="GENERATOR" content="Microsoft FrontPage 6.0"> | 23:21 |
| rm_you | <meta name="GENERATOR" content="Microsoft FrontPage 6.0"> | 23:22 |
| rm_you | yep | 23:22 |
| rm_you | hehe | 23:22 |
| johnsom_ | I just had to look! | 23:22 |
| rm_you | same | 23:22 |
| sbalukoff | HAHA! | 23:22 |
| xgerman | lol | 23:22 |
| sbalukoff | Oh man, I had no idea. XD | 23:22 |
| *** joeroyall has quit IRC | 23:24 | |
| rm_you | it's ok, it's hidden, because FRAMES | 23:24 |
| rm_you | brb | 23:25 |
| *** mlavalle has quit IRC | 23:47 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!