| *** adam_g has quit IRC | 00:09 | |
| *** adam_g has joined #akanda | 00:10 | |
| *** adam_g has quit IRC | 00:10 | |
| *** adam_g has joined #akanda | 00:10 | |
| *** t0mb0 has joined #akanda | 01:14 | |
| *** puranamr has quit IRC | 01:35 | |
| *** puranamr has joined #akanda | 03:46 | |
| *** puranamr has quit IRC | 04:21 | |
| *** puranamr has joined #akanda | 04:35 | |
| *** puranamr has quit IRC | 04:38 | |
| *** puranamr has joined #akanda | 04:39 | |
| *** puranamr has quit IRC | 04:55 | |
| *** t0mb0 has quit IRC | 08:57 | |
| *** puranamr has joined #akanda | 13:22 | |
| *** puranamr has quit IRC | 15:19 | |
| *** openstackgerrit_ has joined #akanda | 15:54 | |
| *** puranamr has joined #akanda | 15:55 | |
| *** puranamr has quit IRC | 16:09 | |
| *** cleverdevil has joined #akanda | 16:17 | |
| *** puranamr has joined #akanda | 16:18 | |
| *** puranamr has quit IRC | 16:43 | |
| *** puranamr has joined #akanda | 16:58 | |
| *** cleverde_ has joined #akanda | 17:57 | |
| *** cleverdevil has quit IRC | 17:59 | |
| *** openstackgerrit_ has quit IRC | 18:07 | |
| openstackgerrit | Merged stackforge/akanda-neutron: Update extension registration for liberty, bump to 2015.2 https://review.openstack.org/208052 | 18:38 |
|---|---|---|
| *** cleverde_ has quit IRC | 19:34 | |
| *** cleverdevil has joined #akanda | 19:35 | |
| openstackgerrit | Merged stackforge/akanda-appliance: Fix for mock API change, bump setup.cfg version to 2015.2 https://review.openstack.org/207697 | 19:58 |
| *** cleverde_ has joined #akanda | 20:10 | |
| *** cleverdevil has quit IRC | 20:10 | |
| *** puranamr has quit IRC | 20:15 | |
| openstackgerrit | Merged stackforge/akanda-rug: Fix debug and browse config issues https://review.openstack.org/204748 | 20:17 |
| openstackgerrit | Merged stackforge/akanda-rug: Move AKANDA_RUG_DIR from settings to plugin.sh, allow overrides https://review.openstack.org/203893 | 20:17 |
| adam_g | holy smokes stuff is merging | 20:18 |
| *** puranamr has joined #akanda | 20:22 | |
| *** puranamr has quit IRC | 20:27 | |
| *** puranamr has joined #akanda | 20:27 | |
| *** cleverde_ is now known as cleverdevil | 21:17 | |
| *** puranamr has quit IRC | 21:28 | |
| *** puranamr has joined #akanda | 21:41 | |
| adam_g | drwahl, do you guys have a way of pushing MTU settings to your router VMs? | 22:57 |
| drwahl | sort of | 23:00 |
| drwahl | i know we can build a new appliance and redeploy them | 23:00 |
| drwahl | but i'm not sure we have a way to do it live | 23:00 |
| adam_g | drwahl, so for vxlan ive found (at least locally) we need to get appliance VMs configured for 1450 MTU to match host tap devices | 23:07 |
| adam_g | drwahl, not sure the best way to achieve that. | 23:07 |
| drwahl | ya, for us, it would be build a new appliance with that MTU set and delete all currently running routers (allowing the rug to deploy new ones) | 23:08 |
| adam_g | drwahl, aren't the interfaces in the VM configured later via SLAAC? | 23:08 |
| *** fzylogic has joined #akanda | 23:10 | |
| drwahl | fzylogic: (04:08:49 PM) adam_g: drwahl, aren't the interfaces in the VM configured later via SLAAC? | 23:10 |
| adam_g | i know neutron supports configuring an pushing MTU settings out via dhcp but not sure how SLAAC works in that regard | 23:11 |
| drwahl | fzylogic has more experience in this realm. hoping he can shed some light | 23:11 |
| fzylogic | I’m checking the appliance code again, but I thought it was configured statically | 23:12 |
| adam_g | k | 23:12 |
| adam_g | the appliance itself doesnt set interfaces, it just lists them AFAICS | 23:17 |
| adam_g | drwahl, anyway, this is the last sticking point for me.. after manually setting the correct MTU on the appliance, the router comes up and everything is happy | 23:18 |
| fzylogic | adam_g: you’re right, it gets the MTU from the OS; it never sets it | 23:19 |
| fzylogic | I don’t see anywhere that neutron lets you tune that, though... | 23:19 |
| fzylogic | doesn’t seem to be an option for networks, subnets, or ports (at least in Juno) | 23:20 |
| markmcclain | fzylogic: yeah mtu settings are in kilo | 23:31 |
| markmcclain | but that will only advert it via dhcp | 23:31 |
| markmcclain | I don't think config drive pushes it | 23:31 |
| fzylogic | I don’t know what spawned this conversation, but it seems to me that this could be something the rug did for router interfaces. | 23:32 |
| markmcclain | adam_g: so we should be able to use a large mtu size on teh network | 23:32 |
| markmcclain | the underlying physical net has jumbo frames enabled right? | 23:33 |
| fzylogic | DHC does with the exception of the public side | 23:33 |
| fzylogic | preeeeety sure we never pushed it that far up | 23:34 |
| adam_g | markmcclain, the mtu of the tap devices is set by neutron, ya? | 23:39 |
| adam_g | puranamr suggested setting the physical NIC to 1550 so that the resulting neutron-managed taps get a 1500 and match the default in guest | 23:39 |
| markmcclain | tap devices are created by nova | 23:39 |
| adam_g | ah right | 23:40 |
| markmcclain | http://docs.openstack.org/juno/config-reference/content/networking-options-plugins-ml2.html | 23:42 |
| markmcclain | see the first tip under the configuration options | 23:43 |
| adam_g | hmm okay ya | 23:44 |
| adam_g | so set the physical interface MTU to +50 the desired | 23:44 |
| markmcclain | we could also just look at what the fix in kilo was | 23:45 |
| markmcclain | and backport enough of it | 23:45 |
| adam_g | i only have experience WRT dhcp | 23:45 |
| adam_g | no idea how/if SLAAC implements something similar | 23:46 |
| markmcclain | so with IPv6 the path len is learned | 23:47 |
| markmcclain | we can send out the MTU via v6 RA, but we're currently not using RA on the mgt net | 23:48 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!