*** cleverdevil has quit IRC | 00:02 | |
*** puranamr has quit IRC | 00:19 | |
*** puranamr has joined #akanda | 00:33 | |
*** puranamr has quit IRC | 01:08 | |
*** t0mb0 has joined #akanda | 01:19 | |
*** puranamr has joined #akanda | 01:29 | |
*** puranamr has quit IRC | 01:33 | |
*** t0mb0 has quit IRC | 01:35 | |
*** t0mb0 has joined #akanda | 02:03 | |
*** puranamr has joined #akanda | 02:10 | |
*** puranamr has quit IRC | 02:15 | |
*** puranamr has joined #akanda | 02:22 | |
*** puranamr has quit IRC | 02:27 | |
*** puranamr has joined #akanda | 05:45 | |
*** puranamr has quit IRC | 05:49 | |
*** openstackgerrit_ has joined #akanda | 09:53 | |
*** ronis_ has joined #akanda | 11:09 | |
*** openstackgerrit has quit IRC | 11:31 | |
*** openstackgerrit_ is now known as openstackgerrit | 11:31 | |
*** openstackgerrit has quit IRC | 11:32 | |
*** openstackgerrit_ has joined #akanda | 11:32 | |
*** openstackgerrit_ is now known as openstackgerrit | 11:32 | |
*** openstackgerrit_ has joined #akanda | 11:33 | |
*** openstackgerrit has quit IRC | 12:16 | |
*** openstackgerrit has joined #akanda | 12:17 | |
*** _t0mb0_ has joined #akanda | 13:10 | |
*** _t0mb0_ has quit IRC | 13:18 | |
*** puranamr has joined #akanda | 13:33 | |
*** t0mb0 has quit IRC | 14:07 | |
*** ronis_ has quit IRC | 15:17 | |
*** cleverdevil has joined #akanda | 16:24 | |
*** openstackgerrit_ has quit IRC | 16:29 | |
*** ronis_ has joined #akanda | 16:32 | |
*** cleverde_ has joined #akanda | 18:32 | |
*** cleverdevil has quit IRC | 18:35 | |
*** ryanpetrello has quit IRC | 19:01 | |
adam_g | https://git.openstack.org/cgit/stackforge/akanda-rug/tree/akanda/rug/tenant.py#n118 | 19:19 |
---|---|---|
adam_g | markmcclain, do you happen to know in which circumstances we'd be getting a message with no router_id and hitting that lookup? i was triggering that a while ago but seem to have forgotten what was causing it | 19:19 |
markmcclain | it could be the router has been deleted but the events are still being processed in the queue | 19:21 |
*** ryanpetrello has joined #akanda | 19:29 | |
adam_g | hm | 19:29 |
openstackgerrit | David Lenwell proposed stackforge/akanda: driver spec https://review.openstack.org/209232 | 19:42 |
adam_g | https://bugs.launchpad.net/akanda/+bug/1481463 | 19:43 |
openstack | Launchpad bug 1481463 in akanda "cant delete akanda ports" [Undecided,New] | 19:43 |
*** cleverde_ has quit IRC | 19:45 | |
*** cleverdevil has joined #akanda | 19:47 | |
*** cleverdevil has quit IRC | 19:51 | |
*** cleverdevil has joined #akanda | 20:02 | |
*** ronis_ has quit IRC | 20:25 | |
drwahl | so i got multi-node devstack up with akanda/vxlan. but the router seems to keep getting stuck in 'build' state even thoug the nova VM is up/happy | 21:15 |
drwahl | any ideas? | 21:15 |
adam_g | drwahl, thats probablyt he MTU | 21:19 |
drwahl | ah ok | 21:19 |
adam_g | drwahl, if you login to the appliance and bump it down to 1450 (assuming the host taps are created 1450) you should be okay | 21:20 |
drwahl | just ssh to the fdca address? | 21:20 |
drwahl | seems to be timing out... | 21:21 |
adam_g | drwahl, ssh isnt going to work till the MTU is adjusted | 21:21 |
adam_g | drwahl, i had to login via console | 21:21 |
adam_g | drwahl, but first i had to mount the appliance image and add my own sudo'd user, i dont know the default root passwdon that image | 21:21 |
drwahl | i might have it somewhere... | 21:22 |
adam_g | drwahl, as for manually hacking in the user: sudo modprobe nbd; sudo qemu-nbd -c /dev/nbd1 ~/akanda.qcow; sudo mount /dev/nbd1p1 /mnt; sudo chroot /mnt; adduser akanda/visudo/etc; sudo umount /mnt; sudo qemu-nbd -d /dev/nbd1 | 21:23 |
adam_g | then republish to glance, update rug.ini | 21:23 |
ryanpetrello | oh holy god | 21:23 |
ryanpetrello | we found the worst akanda-appliance bug ever | 21:23 |
ryanpetrello | so a customer of ours was adding a bunch of networks | 21:23 |
ryanpetrello | and he noticed after he got a certain amount | 21:23 |
ryanpetrello | his router wouldn't boot | 21:24 |
ryanpetrello | so I go diggiong | 21:24 |
ryanpetrello | and I noticed that after I add a *ninth interface* to the router | 21:24 |
ryanpetrello | it tries to hotplug and gets stuck in this bizarre loop | 21:24 |
ryanpetrello | so I go digging | 21:24 |
ryanpetrello | and the API on the router is returning invalid information about the interfaces | 21:24 |
ryanpetrello | namely, if there are more than 8 interfaces, it truncates them at 8 | 21:24 |
ryanpetrello | https://github.com/dreamhost/akanda-appliance/blob/master/akanda/router/drivers/ip.py#L502 | 21:25 |
ryanpetrello | re.split('(^|\n)(?=[0-9]: \w+\d{0,3}:)', data, re.M) | 21:25 |
ryanpetrello | that re.M needs to be `flags=re.M` | 21:25 |
ryanpetrello | because the signature of re.split is re.split(pattern, string, maxsplit=0, flags=0) | 21:25 |
ryanpetrello | so as you've probably guessed, re.M == 8 | 21:25 |
adam_g | markmcclain, historically what was responsible for creating the backing schema for akanda.neutron.db.models_v2.* ? | 21:26 |
markmcclain | nothing.. it was manually added with some hacks | 21:27 |
markmcclain | there's a bug floating around to remove the old models since they aren't used for anything right now | 21:27 |
adam_g | markmcclain, meaning they never actually get created or used? | 21:28 |
markmcclain | they were probably last used grizzly | 21:28 |
adam_g | hm | 21:28 |
adam_g | what about the quota extensions in general/ | 21:28 |
markmcclain | then we killed the feature because it was non-standard | 21:28 |
markmcclain | but left the plumbing for when we returned | 21:28 |
adam_g | the new quota stuff pushed into neutron ends up activating them in a way that they're now active and trying to hit the DB | 21:29 |
adam_g | https://bugs.launchpad.net/akanda/+bug/1481463 | 21:29 |
openstack | Launchpad bug 1481463 in akanda "cant delete akanda ports" [Undecided,New] | 21:29 |
markmcclain | now we're paying the tax for leaving it in place | 21:29 |
markmcclain | we could just remove the non-standard models and add them back later | 21:29 |
adam_g | markmcclain, are the actual quota resources expected to be tracked and functional at this point? | 21:29 |
adam_g | like, all the stuff that gets registered via _authzbase.py | 21:30 |
markmcclain | as now... we're not adding any external resources so none | 21:30 |
* adam_g == confused | 21:32 | |
adam_g | markmcclain, are you saying all of akanda.neutron.extensions.* is bitrot and hasnt been used since grizzly ? | 21:35 |
adam_g | markmcclain, sorry didnt realize you were TC'ing | 21:36 |
markmcclain | yes... significant bitrot in that dir | 21:39 |
markmcclain | https://github.com/stackforge/akanda-neutron/blob/master/akanda/neutron/extensions/routerstatus.py | 21:39 |
markmcclain | is the only thing left in use currently | 21:39 |
adam_g | ok | 21:40 |
markmcclain | now the question is which system that ticket existed in | 21:42 |
openstackgerrit | Ryan Petrello proposed stackforge/akanda-appliance: Don't truncate interface parsing past the ninth interface. https://review.openstack.org/209265 | 21:46 |
ryanpetrello | markmcclain adam_g ^ explains the bug I found a bit better than my rambling above | 21:48 |
ryanpetrello | needless to say, it's...an entertaining one | 21:48 |
adam_g | ryanpetrello, ya, sorry i didnt respond. makes sense what you 'splained | 21:49 |
ryanpetrello | ya, super gross | 21:49 |
adam_g | but what kind of weirdo needs 9 interfaces? come on | 21:49 |
ryanpetrello | haha | 21:49 |
ryanpetrello | +1 | 21:49 |
openstackgerrit | Adam Gandelman proposed stackforge/akanda-neutron: Drop unused Neutron extensions https://review.openstack.org/209267 | 21:51 |
markmcclain | adam_g: should we also drop authz extension and the models? | 22:00 |
adam_g | markmcclain, oh right. models, yes. authz still gets used for regsitraton now. let me see if the rest is needed | 22:00 |
markmcclain | ok | 22:00 |
openstackgerrit | Adam Gandelman proposed stackforge/akanda-neutron: Drop unused Neutron extensions https://review.openstack.org/209267 | 22:08 |
adam_g | markmcclain, stuff in routerstatus.py inherits from the authz classes. maybe it could go away entire, but i dont know enough about how this works to really tell | 22:09 |
* adam_g goes to coffee | 22:10 | |
*** cleverdevil has quit IRC | 22:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!