*** cleverdevil has quit IRC | 00:02 | |
*** adam_g has quit IRC | 00:32 | |
*** adam_g has joined #akanda | 00:32 | |
*** t0mb0 has joined #akanda | 00:50 | |
adam_g | drwahl, http://paste.openstack.org/show/393505/ dont suppose you've hit this in getting stable/juno up? | 00:55 |
---|---|---|
adam_g | oh, nvm | 01:05 |
openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Move AKANDA_RUG_DIR from settings to plugin.sh, allow overrides https://review.openstack.org/203893 | 01:11 |
adam_g | so i needed this to get stable/juno deployed okay https://review.openstack.org/#/c/203897/ | 01:34 |
adam_g | drwahl, that gets me an instance booted but cannot be reached, dont seem to hit the same issue as you | 01:34 |
adam_g | s/instance/router instance/ | 01:52 |
*** t0mb0 has quit IRC | 02:01 | |
*** t0mb0 has joined #akanda | 02:04 | |
*** fzylogic_ has joined #akanda | 05:48 | |
*** fzylogic has quit IRC | 05:49 | |
*** fzylogic_ is now known as fzylogic | 05:49 | |
*** t0mb0 has quit IRC | 12:53 | |
*** markmcclain has quit IRC | 15:09 | |
*** ryanpetrello has quit IRC | 15:09 | |
*** sarob has quit IRC | 15:09 | |
*** puranamr has joined #akanda | 16:07 | |
*** puranamr has quit IRC | 16:38 | |
*** puranamr has joined #akanda | 16:40 | |
puranamr | can't join webinar? it is cribbing for adobe flash! | 17:27 |
*** markmcclain has joined #akanda | 17:27 | |
adam_g | unit tests failing and blocking merges, fixed by https://review.openstack.org/#/c/203819/ plz review | 17:32 |
adam_g | davidlenwell_, around? | 17:33 |
davidlenwell_ | yeah | 17:33 |
*** davidlenwell_ is now known as davidlenwell | 17:33 | |
*** ChanServ sets mode: +o davidlenwell | 17:33 | |
davidlenwell | sup? | 17:34 |
adam_g | davidlenwell, deploying vxlan multihost--is the router appliance expected to be reachable before the 2nd node is installed and vxfld setup? | 17:41 |
davidlenwell | adam_g: good questions.. I'm not sure .. sean and rama picked up that project and got vxfld working while I was put on something else.. | 17:42 |
davidlenwell | puranamr: ^^ ? | 17:42 |
adam_g | davidlenwell, ah | 17:42 |
adam_g | davidlenwell, also, not sure if you saw my msg. but your crazy osx editors seem to be saving files yo touch with 0755 permissions | 17:43 |
puranamr | yes | 17:43 |
adam_g | nose apparently skips test_*.py files that are executable | 17:43 |
adam_g | so thats why our devstack test hasn't been blowing up on failures recently, t hasn't been running | 17:43 |
davidlenwell | adam_g: lame.. I was trying a new editor on seans recomendation .. looks like im gonna go back to vi in terminal windows. | 17:44 |
puranamr | good find adam_g! | 17:44 |
adam_g | davidlenwell, im sure its configurable. which editor? atom? | 17:44 |
davidlenwell | pycharm | 17:44 |
davidlenwell | it probably is | 17:44 |
davidlenwell | it wasn't consistant .. so im not 100% sure it was the editor | 17:45 |
adam_g | ah | 17:45 |
davidlenwell | I'll look into it | 17:46 |
davidlenwell | thanks for digging into that | 17:46 |
adam_g | https://review.openstack.org/#/c/204185/ and https://review.openstack.org/#/c/204188/ fix unit tests similarly on stable branches | 17:56 |
*** sarob has joined #akanda | 18:14 | |
*** ryanpetrello has joined #akanda | 18:16 | |
sarob | adam_g: better to just use kilo nova and akanda for now to build demos | 18:21 |
sarob | adam_g: dont you think? | 18:21 |
adam_g | sarob, maybe. does kilo still build? :p | 18:28 |
adam_g | the liberty 1 milestone should be fine | 18:28 |
adam_g | the nova patch in question merged like last week i think | 18:29 |
sarob | adam_g: i was focused on getting stable/juno configs completed | 18:31 |
sarob | adam_g: last time i tried building kilo it failed | 18:31 |
adam_g | sarob, the keystone issue you guys were hitting is user error | 18:31 |
sarob | adam_g: whose? :) | 18:31 |
adam_g | dunno? | 18:32 |
sarob | adam_g: whats the errororo | 18:32 |
adam_g | the HOST_IP in localrc was set to the wrong IP | 18:32 |
adam_g | so keystone was trying and failing to bind to a non-existent addr | 18:32 |
sarob | adam_g: son of a nutcracker | 18:32 |
davidlenwell | doh | 18:32 |
adam_g | in the future if things fail to start in devstack, try running the thing manually to get some debug infos | 18:32 |
adam_g | keystone is sorta special case because devstack sets it up behind mod_wsgi, so something like that isn't immediately obvious in apache logs | 18:33 |
sarob | meaning service | 18:33 |
sarob | right | 18:33 |
sarob | ? | 18:33 |
adam_g | yea | 18:33 |
adam_g | you can run keystone outside of apache by running keystone-all | 18:33 |
adam_g | other services shoudl spit the failures into screen session | 18:33 |
sarob | adam_g: yeah | 18:34 |
sarob | ill take another stab and double check that my juno fail was for that reason | 18:35 |
adam_g | davidlenwell, sshuttle -r adam@192.168.25.30 10.0.0.0/24 | 18:36 |
davidlenwell | yeah .. i've already been playing with sshuttle | 18:37 |
adam_g | davidlenwell, gets me a janky vpn to everything on 10.0.0.0/24 behind 192.168.25.30 via ssh | 18:37 |
adam_g | cool | 18:37 |
davidlenwell | its awsome | 18:37 |
adam_g | yea, super handy. i run it in a screen/tmux session in the foreground tho. its daemon mode is kinda wonky and sometimes doesn't clean iptables up properly on shutdown | 18:37 |
sarob | davidlenwell and adam_g get us to the point of being able to demo this friday | 18:43 |
sarob | https://docs.google.com/a/akanda.io/presentation/d/1u3zjge0mWq3uskmlZY2QSzISQWh7ZEnuQmJLMo70o0E/edit?usp=sharing | 18:43 |
adam_g | sarob, all you need is a snapshotted devstack ? | 18:45 |
sarob | yup | 18:45 |
sarob | i have a bunch of other stuff to do. | 18:46 |
sarob | y'all picking this up would be super duper | 18:46 |
davidlenwell | we god it | 18:52 |
davidlenwell | got it .. I'm also picking back up on the driver stuff | 18:52 |
sarob | y'all putting the demo config into the ops repo so i can recreate when required | 18:55 |
sarob | puranamr: i need you to get stable/juno vxlan akanda configuration into the ops repo | 18:57 |
puranamr | ok | 18:57 |
sarob | puranamr: verify whats there is good is more of what i meant | 18:58 |
puranamr | yeah understood | 18:58 |
sarob | puranamr: work with drwahl and rods if they are still having problems with the configs | 18:59 |
sarob | puranamr: cool thx | 18:59 |
puranamr | ok | 18:59 |
sarob | puranamr: foot feeling better? | 18:59 |
puranamr | i was rushing to get to the team meeting and put a wrong step out and sprained the ankle, so had to retreat back :( | 19:00 |
puranamr | thanks for asking! | 19:00 |
sarob | puranamr: feel better. meetings arent worth getting injured for ;) | 19:04 |
puranamr | :) | 19:05 |
drwahl | sarob: after running the vxfld services, i was kind of expecting there to be some sort of briding between the controller node and the compute node | 19:32 |
drwahl | but there was none. am i mistaken in expecting to see some sort of bridge or interface? | 19:32 |
puranamr | drwahl: i can work with you helping set the vxfld service | 19:42 |
drwahl | it seemed like they were easy to get them working given the instructions i followed here: https://docs.google.com/document/d/1xSjYVUO6J88QS6nah1TBBLHBm8iIcXjr_9t8SkMxovw/edit# | 19:42 |
puranamr | that link is asking for permission to access | 19:44 |
drwahl | i think sarob created that document | 19:45 |
drwahl | but it essentially says to run the typical 'python setup.py install', then muck with 1 config file, then run vxsnd/vxrd (depending on which node) | 19:46 |
sarob | puranamr: use your akanda account | 19:46 |
puranamr | sarob: ok | 19:46 |
puranamr | no you wouldn't see any br instances in the running vm instances. | 19:51 |
puranamr | u will see in the host, like vxlan-1009 etc | 19:51 |
*** cleverdevil has joined #akanda | 19:51 | |
puranamr | based upon if it is controller / compute | 19:51 |
drwahl | right, but i'm not seeing that on the controller (or on the compute node) | 19:57 |
drwahl | will those interfaces be created shortly after the vxfld services are started? | 19:57 |
puranamr | no when you run the devstack stack.sh script it gets created | 20:01 |
puranamr | the service is started after the completion of the stack.sh script | 20:01 |
puranamr | to compare notes, u could log onto 67.50.59.148 dhc-user to see the interface list | 20:02 |
drwahl | ah, that'll be helpful. thanks | 20:02 |
puranamr | sorry 67.205.59.148 | 20:03 |
drwahl | bleh, looks like it's expecting an ssh key that i don't have | 20:04 |
puranamr | u could run ssh-import-id drwahl and the try connecting | 20:05 |
drwahl | i have my keys loaded. i'm not sure how that server would have my public key though | 20:06 |
puranamr | try connecting | 20:06 |
drwahl | nope. pubkey denied | 20:07 |
puranamr | did you use dhc-user@67.205.59.148? | 20:07 |
puranamr | did it let in? | 20:08 |
drwahl | http://pastebin.com/ZH5u06iE | 20:08 |
puranamr | can u run with -vvv option and paste the output | 20:09 |
drwahl | http://pastebin.com/0hB50SDy | 20:10 |
adam_g | puranamr, did you add his pub key to dhc-user's ~/.ssh/authorized_keys ? | 20:11 |
puranamr | adam_g: no | 20:12 |
adam_g | puranamr, the ssh-import-id thing only works for users who have launchpad accounts /w ssh keys loaded there | 20:12 |
puranamr | adam_g: ok i see | 20:12 |
drwahl | ya. wasn't sure how my pub key would be on that box :) | 20:12 |
puranamr | adam_g: can i go ahead and add his public key to the authorized_keys list | 20:14 |
drwahl | ok, i think i may have found at least part of the issue i'm having: we had to install vxfld into a virtualenv because the 'setup.py install' command was breaking our pythonpath | 20:15 |
puranamr | drwahl: can u put in ur public key | 20:15 |
puranamr | ok | 20:15 |
adam_g | puranamr, you could do whatever you want, its not my system. :) | 20:16 |
drwahl | http://pastebin.com/yg9jNu9E | 20:16 |
puranamr | adam_g: :) | 20:16 |
drwahl | so after stack.sh finishes, the vxfld services are not running, which suggests to me that stack.sh isn't aware of them | 20:16 |
puranamr | yes vxfld is outside stack.sh | 20:17 |
drwahl | oh ok. when you said "the services gets started after stack.sh", i thought you meant as part of the stack.sh scripts | 20:18 |
drwahl | but you must mean manually | 20:18 |
drwahl | correct? | 20:18 |
puranamr | correct | 20:18 |
drwahl | ok, so after stack.sh, i don't see any vxlan interface, regardless of whether i start up the vxfld services or not | 20:19 |
drwahl | so maybe my localrc is messed up? | 20:19 |
puranamr | possible | 20:20 |
drwahl | http://pastebin.com/Tfp3AQph | 20:20 |
drwahl | that's the controller node | 20:20 |
puranamr | why is the ak-rug service enabled? | 20:22 |
puranamr | i see that the conf that was shared has it | 20:23 |
puranamr | not your fault :) | 20:23 |
drwahl | sarob: ^^? | 20:23 |
sarob | drwahl: asking why ak-rug enabled in local.conf? | 20:24 |
*** cleverdevil has quit IRC | 20:24 | |
drwahl | ya | 20:24 |
puranamr | drwahl: can u pl unstack and comment the line enable_service ak-rug | 20:24 |
drwahl | sure | 20:24 |
puranamr | after u unstack, pl see ifconfig -a / ip link show to ensure that you do not see any interfaces hanging around | 20:25 |
drwahl | there are a ton of tap interfaces | 20:26 |
drwahl | after the unstack | 20:27 |
drwahl | i'll just rebuild, since i have this documented/automated with vagrant | 20:28 |
puranamr | ok, before you run stack.sh, pl comment out the line enable_service ak-rug | 20:31 |
drwahl | will do. i'll let you know when stack.sh completes | 20:32 |
puranamr | sure | 20:35 |
openstackgerrit | Merged stackforge/akanda-rug: Remove remaining references to oslo.cfg namespace https://review.openstack.org/203819 | 20:52 |
drwahl | ok, stack.sh finally finished | 21:10 |
drwahl | and all i have is eth0, lo and virbr0 | 21:10 |
puranamr | on the control nodde? | 21:15 |
drwahl | on both, fwiw | 21:15 |
puranamr | something doesn't add up | 21:16 |
puranamr | devstack log? | 21:16 |
drwahl | incoming.... | 21:16 |
adam_g | puranamr, huh? why should he be removing 'enable_service ak-rug' ? | 21:17 |
puranamr | adam_g: this is to bring up the vxfld (software vteps) in his lab environment. | 21:18 |
drwahl | https://gist.github.com/anonymous/d2c6626d6b168f289582 | 21:18 |
adam_g | puranamr, right, but that gets stood up alongside akanda, no? | 21:18 |
puranamr | this is being done on stable juno | 21:19 |
adam_g | ? | 21:20 |
puranamr | drwahl: looking at the log, few cycles ... | 21:22 |
puranamr | drwahl: did you follow all the steps outlined in the doc, without missing anything | 21:29 |
drwahl | here's my setup script: http://pastebin.com/VqRp3UKn | 21:30 |
drwahl | it was in-line with the document at one point | 21:31 |
puranamr | i do not see apt-get upgrade after update | 21:32 |
drwahl | i can give that a shot. i tried it earlier and it didnt' seem to make much of a difference. our images usually aren't *too* far out of date | 21:34 |
drwahl | but i'll give it a shot to rule it out | 21:34 |
puranamr | one more thing, | 21:37 |
puranamr | i am comparing notes with the config that i used. so few seconds more | 21:42 |
puranamr | drwahl: can u sent the content of controller_local.conf? | 21:45 |
puranamr | s/sent/send | 21:45 |
drwahl | yup, one moment | 21:45 |
drwahl | http://pastebin.com/GJUqn6T1 | 21:46 |
puranamr | pl comment out ak-rug and uncomment q-dhcp, q-l3 and q-meta | 21:48 |
puranamr | rest is fine | 21:48 |
puranamr | similarly the compute_local.conf too | 21:48 |
puranamr | in pastebin | 21:48 |
drwahl | http://pastebin.com/6yfs9NVJ | 21:49 |
drwahl | restacking with those changes | 21:51 |
*** fzylogic has quit IRC | 21:51 | |
puranamr | one more minute | 21:52 |
puranamr | pl change the NOVNCPROXY_URL to the SERVICE_HOST IP | 21:53 |
puranamr | it is referring to the example values still | 21:53 |
*** fzylogic has joined #akanda | 21:53 | |
drwahl | ah, ya | 21:53 |
puranamr | and i see couple of plugin info also missing in the compute local.conf | 21:53 |
drwahl | we're really just wanting to see the vxlan stuff in action and get a setup where we can begin playing with it | 21:54 |
drwahl | so the vnc stuff probably isn't a huge deal | 21:54 |
puranamr | not vnc | 21:54 |
puranamr | but plugins are missing | 21:54 |
puranamr | Q_ML2_PLUGIN_MECHANISM_DRIVERS=linuxbridge | 21:56 |
puranamr | Q_ML2_TENANT_NETWORK_TYPE=vxlan | 21:56 |
puranamr | rest is ok | 21:57 |
drwahl | that needs to be in the compute nodes local.conf? | 21:57 |
puranamr | yes correct | 21:57 |
puranamr | sorry my bad, just a sec | 21:57 |
puranamr | no pl ignore, multiple screens here | 21:58 |
puranamr | got messed up | 21:58 |
puranamr | sorry for the confusion. u r ok with the compute | 21:59 |
drwahl | sorry, just to be clear, should the compute local.conf have those plugins or not? | 21:59 |
puranamr | controller should have | 22:01 |
puranamr | not the compute | 22:01 |
drwahl | ok | 22:01 |
drwahl | i'm just about done for the day. i'm going to rebuild these VMs from script (vagrant) and kick off a stack.sh with these localrc changes and then take off | 22:04 |
drwahl | i'll touch bases with you tomorrow on my progress | 22:04 |
drwahl | thanks for your help today puranamr | 22:04 |
puranamr | sure np, still comparing the devstack.log. i will be around to help you, if you will be around | 22:04 |
puranamr | oh sorry did not see the earlier msg | 22:05 |
puranamr | sure thanks | 22:05 |
*** cleverdevil has joined #akanda | 22:44 | |
*** puranamr has quit IRC | 22:56 | |
*** puranamr has joined #akanda | 23:08 | |
*** cleverdevil has quit IRC | 23:37 | |
*** puranamr has quit IRC | 23:41 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!