| *** markmcclain has quit IRC | 00:53 | |
| *** markmcclain has joined #openstack-astara | 00:55 | |
| *** cleverdevil has quit IRC | 01:09 | |
| *** markmcclain has quit IRC | 02:39 | |
| *** markmcclain has joined #openstack-astara | 02:40 | |
| openstackgerrit | Yang Li proposed openstack/astara: Remove 'FIXME' if it's alreasy be done https://review.openstack.org/266643 | 02:47 |
|---|---|---|
| *** markmcclain has quit IRC | 03:12 | |
| *** markmcclain has joined #openstack-astara | 03:15 | |
| *** markmcclain has quit IRC | 03:37 | |
| *** markmcclain has joined #openstack-astara | 03:38 | |
| openstackgerrit | Adam Gandelman proposed openstack/astara: fail func tests early (do not merge) https://review.openstack.org/266666 | 03:52 |
| *** markmcclain has quit IRC | 03:52 | |
| *** markmcclain has joined #openstack-astara | 04:24 | |
| *** markmcclain has quit IRC | 04:29 | |
| *** markmcclain has joined #openstack-astara | 05:12 | |
| *** markmcclain has quit IRC | 05:21 | |
| *** markmcclain has joined #openstack-astara | 05:50 | |
| *** markmcclain has quit IRC | 05:56 | |
| openstackgerrit | Yang Hongyang proposed openstack/astara: Refactor ensure_cache for instance manager https://review.openstack.org/266740 | 07:27 |
| openstackgerrit | Yang Hongyang proposed openstack/astara: Refactor ensure_cache for loadbalancer driver https://review.openstack.org/266741 | 07:27 |
| openstackgerrit | Yang Hongyang proposed openstack/astara: Refactor ensure_cache for router driver https://review.openstack.org/266742 | 07:27 |
| openstackgerrit | Yang Hongyang proposed openstack/astara: Allow running individual test using tox https://review.openstack.org/266758 | 08:00 |
| openstackgerrit | Yang Hongyang proposed openstack/astara: Refactor ensure_cache for loadbalancer driver https://review.openstack.org/266741 | 08:15 |
| openstackgerrit | Yang Hongyang proposed openstack/astara: Refactor ensure_cache for instance manager https://review.openstack.org/266740 | 08:15 |
| openstackgerrit | Yang Hongyang proposed openstack/astara: Refactor ensure_cache for router driver https://review.openstack.org/266742 | 08:15 |
| *** ronis has joined #openstack-astara | 08:48 | |
| openstackgerrit | Yang Hongyang proposed openstack/astara: Use assertEqual instead of assertEquals https://review.openstack.org/266804 | 09:26 |
| *** owlbot has quit IRC | 09:30 | |
| *** owlbot has joined #openstack-astara | 09:31 | |
| openstackgerrit | Yang Li proposed openstack/astara: Remove 'FIXME' if it's already be done https://review.openstack.org/266643 | 10:20 |
| openstackgerrit | xiayu proposed openstack/astara: Astara oslo.rootwrap https://review.openstack.org/264759 | 10:47 |
| *** leonstack has quit IRC | 11:06 | |
| *** leonstack has joined #openstack-astara | 11:07 | |
| openstackgerrit | xiayu proposed openstack/astara: Automatically generate etc/orchestrator.ini file https://review.openstack.org/265576 | 12:28 |
| *** davidlenwell has quit IRC | 14:57 | |
| *** davidlenwell has joined #openstack-astara | 14:59 | |
| *** ronis has quit IRC | 16:24 | |
| rods | adam_g looks like https://review.openstack.org/#/c/266551/1 is failing because devstack is not able to create the public network http://logs.openstack.org/51/266551/1/check/gate-functional-dsvm-astara/ef85e22/logs/devstacklog.txt.gz | 16:25 |
| *** cleverdevil has joined #openstack-astara | 17:09 | |
| *** Guest32 has joined #openstack-astara | 17:12 | |
| Guest32 | adam_g: I know the cause of the test failures | 17:25 |
| *** Guest32 is now known as markmcclain1 | 17:25 | |
| *** ronis has joined #openstack-astara | 17:30 | |
| *** cleverdevil has quit IRC | 17:36 | |
| *** leonstack1 has joined #openstack-astara | 18:05 | |
| *** leonstack has quit IRC | 18:07 | |
| adam_g | markmcclain1, whats the cause? | 18:18 |
| adam_g | i noticed that failing late last night | 18:18 |
| markmcclain1 | mordred and I are discussing -dev | 18:18 |
| adam_g | i think this is what i was hitting | 18:19 |
| adam_g | https://review.openstack.org/#/c/263839/ | 18:19 |
| adam_g | (rechecking to find out) | 18:19 |
| *** leonstack has joined #openstack-astara | 18:39 | |
| *** leonstack1 has quit IRC | 18:40 | |
| openstackgerrit | Rosario Di Somma proposed openstack/astara: Delete detached vrrp ports when a newtork is removed https://review.openstack.org/267088 | 18:43 |
| rods | adam_g markmcclain1 ^^ that should give you an idea about what's needed to fix the issue | 18:46 |
| markmcclain1 | rods: awesome | 18:55 |
| *** cleverdevil has joined #openstack-astara | 19:11 | |
| openstackgerrit | Adam Gandelman proposed openstack/astara: Drop unused call to non-existent function https://review.openstack.org/267127 | 19:31 |
| openstackgerrit | Adam Gandelman proposed openstack/astara: Drop astara.openstack and astara.service https://review.openstack.org/267130 | 19:42 |
| *** davidlenwell has quit IRC | 19:57 | |
| *** davidlenwell has joined #openstack-astara | 19:58 | |
| *** stanchan has joined #openstack-astara | 20:05 | |
| markmcclain1 | adam_g: it's like watching paint dry waiting for the functional job to run on your patch | 20:11 |
| adam_g | markmcclain1, http://i.dailymail.co.uk/i/pix/2013/03/27/article-2300124-18F5C407000005DC-552_634x490.jpg | 20:12 |
| markmcclain1 | haha | 20:12 |
| adam_g | markmcclain1, so about HA + managing instances | 20:12 |
| markmcclain1 | yeah | 20:12 |
| adam_g | markmcclain1, my initial plan is to handle everything/most of it in the instance manager | 20:13 |
| adam_g | so that a single instance manager is capable of managing a single or multiple instances | 20:13 |
| adam_g | replacing self.instance_manager with self.instances, which is a resource container similar to tenant.py's state machine ResourceContainer | 20:13 |
| adam_g | self.instances would have the logic for doing things like pushing config/verifying interfaces/etc across multiples | 20:13 |
| adam_g | we need to account for cases where a subset of the managed instances are down/are out of date/etc | 20:14 |
| adam_g | so im thinking a states.DEGRADED be added | 20:14 |
| markmcclain1 | yeah | 20:14 |
| markmcclain1 | that would make sense | 20:14 |
| adam_g | as you mentioned yesterday, update_state and lots of other things in instance manager are really ready for a refactor so this might be a substantial stack of patches | 20:15 |
| markmcclain1 | yeah | 20:16 |
| markmcclain1 | also concerned how the state machine would change | 20:16 |
| markmcclain1 | because checkboot, replug, and rebuild would be handled | 20:16 |
| markmcclain1 | in the HA case for rebuild | 20:17 |
| adam_g | dlchmm | 20:17 |
| markmcclain1 | seems like you want to attempt to rebuild one at a time | 20:18 |
| adam_g | markmcclain1, you being the user, or you being the orchestrator? | 20:19 |
| markmcclain1 | and orchestrator | 20:19 |
| adam_g | id think the operator would want to rebuild all the nodes in a group in one command (ie, to upgrade image) | 20:19 |
| markmcclain1 | but that has some interesting side-effects in state machine | 20:19 |
| adam_g | the orchestrator could stagger it across the nodes for sake of uptime | 20:20 |
| adam_g | markmcclain1, im thinking for things like replug, any exit from replug() that currently ends up in RESTART could also exit in DEGRADED | 20:23 |
| markmcclain1 | yeah.. assuming alive could exit into degraded | 20:23 |
| adam_g | the DEGRADED state would be a new one where we cycle through the managed instances ensuring essentially all the things we do now for singles (the instance is there, plugged, configured, etc) | 20:23 |
| markmcclain1 | am concerned that if one of the pairs get stuck down | 20:24 |
| markmcclain1 | that we're preventing the healthy node from getting current | 20:24 |
| adam_g | markmcclain1, (in my head) that should be non-blocking for the healthy node--it'd just be stuck in a degraded state with only 1/2 routers functional | 20:25 |
| markmcclain1 | ok.. so essentially we're breaking down into machines for each instance | 20:25 |
| markmcclain1 | and the master machine is really for the logical one? | 20:26 |
| adam_g | hmm maybe | 20:26 |
| markmcclain1 | might actually make things simpler | 20:28 |
| *** markmcclain has joined #openstack-astara | 20:38 | |
| markmcclain | fzylogic: much later ping | 20:47 |
| markmcclain | fzylogic: s/ping/pong/ | 20:47 |
| fzylogic | we were having some problems yesterday that seem to suggest the ml2 driver isn't using the floatingip_subnet config option | 20:49 |
| fzylogic | floating IPs were being assigned from an ipv6 subnet | 20:50 |
| markmcclain | fzylogic: urg | 20:50 |
| markmcclain | kilo right? | 20:50 |
| fzylogic | yeah | 20:51 |
| fzylogic | from floatingip-show: | 20:51 |
| fzylogic | | ef032214-3393-4dc3-8d73-5983ca32dab2 | 10.10.10.2 | 2607:f298:5:101d:f816:3eff:fe0c:ecd9 | ad08aae3-ea08-48a9-b5ac-a340bb2c773b | | 20:51 |
| openstackgerrit | Adam Gandelman proposed openstack/astara: gate fixer: Fix devstack auth issue and new PEP8 violation https://review.openstack.org/267165 | 20:55 |
| *** ronis has quit IRC | 21:05 | |
| markmcclain | fzylogic: I think I know what's happening | 21:13 |
| fzylogic | yeah? | 21:13 |
| markmcclain | while the proper class is being mixed in | 21:13 |
| markmcclain | it's not being mixed in the correct place, so REST call is dispatched to the original version not the updated version | 21:14 |
| fzylogic | aha | 21:14 |
| markmcclain | neutron's stable/liberty has some code which kind of does the right thing, but not fully | 21:28 |
| markmcclain | digging deeper will have to test a bit and see if the code needs a bit of modernization | 21:29 |
| fzylogic | ok | 21:29 |
| *** stanchan has quit IRC | 21:29 | |
| markmcclain | this file did not have coverage so tests never exercised it | 21:29 |
| *** stanchan has joined #openstack-astara | 21:30 | |
| markmcclain | also is it still a requirement that floating-ips come out of certain subnets? | 21:30 |
| markmcclain | or can it be simplified to ensure they're just v4 | 21:30 |
| fzylogic | that's a tough one... | 21:32 |
| fzylogic | I think the v4 issue is the larger one we have to deal with | 21:32 |
| fzylogic | the impetus behind wanting to specify specific subnets was so we could retire older ones if we ever needed to reclaim them | 21:32 |
| *** justinlund has joined #openstack-astara | 21:32 | |
| fzylogic | but even with that ability, we never managed to actually retire the old masseffect network after all this time | 21:33 |
| fzylogic | so I think I'd be OK losing that support for now | 21:33 |
| markmcclain | ok | 21:33 |
| fzylogic | it'd mean one less thing for us to maintain | 21:33 |
| markmcclain | seems like updating the allocation pool to be empty could serve as a workaround | 21:34 |
| markmcclain | if you need that functionality again | 21:34 |
| fzylogic | that has its own set of complications, but yes | 21:36 |
| rods | adam_g markmcclain if you find a sec later to leave a comment on https://review.openstack.org/#/c/267088/1 would be great | 22:26 |
| rods | thx | 22:26 |
| rods | see you all tomorrow | 22:26 |
| markmcclain | rods: will do | 22:27 |
| *** markmcclain1 has quit IRC | 22:41 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
