| *** lloydde has quit IRC | 00:02 | |
| *** edygarcia has joined #openstack-dev | 00:03 | |
| *** aeperezt has quit IRC | 00:12 | |
| *** pixelbeat has quit IRC | 00:12 | |
| openstackgerrit | A change was merged to openstack/nova: Support for several HA RabbitMQ servers. https://review.openstack.org/13665 | 00:17 |
|---|---|---|
| *** aeperezt has joined #openstack-dev | 00:22 | |
| *** hemna has quit IRC | 00:27 | |
| *** winston-d has joined #openstack-dev | 00:27 | |
| *** littleidea has joined #openstack-dev | 00:29 | |
| *** littleidea has quit IRC | 00:32 | |
| *** e1mer has joined #openstack-dev | 00:33 | |
| *** littleidea has joined #openstack-dev | 00:37 | |
| *** edygarcia has quit IRC | 00:39 | |
| *** edygarcia has joined #openstack-dev | 00:43 | |
| *** jdurgin has quit IRC | 00:44 | |
| *** markmcclain has quit IRC | 00:46 | |
| *** Mandell has quit IRC | 00:48 | |
| *** Ryan_Lane has quit IRC | 00:52 | |
| *** danwent has joined #openstack-dev | 01:04 | |
| *** zhuadl has quit IRC | 01:05 | |
| *** danwent has quit IRC | 01:06 | |
| *** gongysh_ has quit IRC | 01:09 | |
| *** gongysh_ has joined #openstack-dev | 01:10 | |
| *** edygarcia has quit IRC | 01:17 | |
| *** jtran_ has quit IRC | 01:17 | |
| *** e1mer has quit IRC | 01:20 | |
| *** s0mik has quit IRC | 01:37 | |
| *** PotHix has quit IRC | 01:39 | |
| *** wenjianhn has joined #openstack-dev | 01:39 | |
| *** xchu has joined #openstack-dev | 01:57 | |
| *** soosiechoi has joined #openstack-dev | 02:06 | |
| *** crandquist has joined #openstack-dev | 02:07 | |
| *** dolphm has joined #openstack-dev | 02:10 | |
| *** tgall_foo has joined #openstack-dev | 02:19 | |
| *** tgall_foo has quit IRC | 02:19 | |
| *** tgall_foo has joined #openstack-dev | 02:19 | |
| *** adjohn has quit IRC | 02:22 | |
| *** darjeeling has quit IRC | 02:27 | |
| *** adjohn has joined #openstack-dev | 02:34 | |
| *** adjohn has quit IRC | 02:37 | |
| *** kobier has joined #openstack-dev | 02:38 | |
| *** crandquist has quit IRC | 02:42 | |
| *** waa has quit IRC | 02:46 | |
| *** markmcclain has joined #openstack-dev | 03:04 | |
| *** e1mer has joined #openstack-dev | 03:06 | |
| *** zbitter has joined #openstack-dev | 03:09 | |
| *** kobier has quit IRC | 03:10 | |
| *** gongysh_ has quit IRC | 03:11 | |
| *** zaneb has quit IRC | 03:13 | |
| *** ben_duyujie has joined #openstack-dev | 03:13 | |
| *** danwent has joined #openstack-dev | 03:17 | |
| *** jemartin has quit IRC | 03:24 | |
| *** ben_duyujie has quit IRC | 03:25 | |
| *** torandu has joined #openstack-dev | 03:25 | |
| *** dolphm has quit IRC | 03:27 | |
| *** e1mer has quit IRC | 03:32 | |
| *** \etc\bin has joined #openstack-dev | 03:32 | |
| *** nunosantos has quit IRC | 03:34 | |
| *** nunosantos_ has quit IRC | 03:35 | |
| *** roge has quit IRC | 03:38 | |
| *** novas0x2a|laptop has quit IRC | 03:39 | |
| *** soosiechoi has quit IRC | 03:43 | |
| *** Mandell has joined #openstack-dev | 03:54 | |
| *** markmcclain has quit IRC | 04:01 | |
| *** ben_duyujie has joined #openstack-dev | 04:05 | |
| *** markmcclain has joined #openstack-dev | 04:14 | |
| *** sc68cal has quit IRC | 04:17 | |
| *** crandquist has joined #openstack-dev | 04:21 | |
| *** markmcclain has quit IRC | 04:21 | |
| *** markmcclain1 has joined #openstack-dev | 04:21 | |
| *** ben_duyujie has quit IRC | 04:25 | |
| *** tgall_foo has quit IRC | 04:29 | |
| *** sc68cal has joined #openstack-dev | 04:29 | |
| *** sc68cal has joined #openstack-dev | 04:29 | |
| *** koolhead17 has joined #openstack-dev | 04:32 | |
| *** Mandell has quit IRC | 04:34 | |
| *** s0mik has joined #openstack-dev | 04:37 | |
| *** bencherian has quit IRC | 04:45 | |
| *** dolphm has joined #openstack-dev | 04:47 | |
| *** bencherian has joined #openstack-dev | 04:53 | |
| *** mikal has quit IRC | 04:55 | |
| *** aeperezt has quit IRC | 04:55 | |
| *** mikal has joined #openstack-dev | 04:56 | |
| *** thingee is now known as thingee_zz | 04:58 | |
| *** hattwick has quit IRC | 05:00 | |
| *** EmilienM has joined #openstack-dev | 05:02 | |
| *** markmcclain1 has quit IRC | 05:05 | |
| *** garyk has quit IRC | 05:11 | |
| *** hattwick has joined #openstack-dev | 05:12 | |
| *** crandquist has quit IRC | 05:13 | |
| *** mikal has quit IRC | 05:14 | |
| *** mikal has joined #openstack-dev | 05:15 | |
| *** sniperd has quit IRC | 05:15 | |
| *** sniperd has joined #openstack-dev | 05:15 | |
| *** sniperd has quit IRC | 05:29 | |
| *** Mandell has joined #openstack-dev | 05:31 | |
| *** adalbas has quit IRC | 05:36 | |
| *** crandquist has joined #openstack-dev | 05:45 | |
| *** adalbas has joined #openstack-dev | 05:49 | |
| rmk | This is a fun one. Scheduler allows you to overcommit memory and is set to 1.5 by default. However, the live migration memory check code path doesn't honor that at all. | 05:51 |
| rmk | So as soon as you're over 1.0, you cannot live migrate to that destination anymore. | 05:51 |
| *** s0mik has quit IRC | 05:52 | |
| *** garyk has joined #openstack-dev | 05:59 | |
| *** littleidea has quit IRC | 06:02 | |
| * ttx loves the smell of the empty zuul gate in the release day morning | 06:05 | |
| *** mindpixel has joined #openstack-dev | 06:05 | |
| *** zhuadl has joined #openstack-dev | 06:11 | |
| *** mrunge has joined #openstack-dev | 06:11 | |
| *** adjohn has joined #openstack-dev | 06:13 | |
| *** xchu has quit IRC | 06:13 | |
| *** alobbs has joined #openstack-dev | 06:24 | |
| *** adjohn has quit IRC | 06:24 | |
| *** winston-d has quit IRC | 06:29 | |
| *** koolhead17 has quit IRC | 06:34 | |
| *** EmilienM has quit IRC | 06:42 | |
| *** eglynn has quit IRC | 06:45 | |
| *** zhuadl has quit IRC | 06:46 | |
| *** zhuadl has joined #openstack-dev | 06:47 | |
| *** zhuadl has joined #openstack-dev | 06:49 | |
| *** epim has joined #openstack-dev | 06:49 | |
| *** rafaduran has joined #openstack-dev | 06:49 | |
| *** spn has joined #openstack-dev | 06:51 | |
| *** eglynn has joined #openstack-dev | 07:03 | |
| *** mattstep has quit IRC | 07:14 | |
| *** mattstep has joined #openstack-dev | 07:14 | |
| *** mattstep has quit IRC | 07:15 | |
| *** mattstep has joined #openstack-dev | 07:16 | |
| *** mattstep has joined #openstack-dev | 07:16 | |
| *** reidrac has joined #openstack-dev | 07:20 | |
| *** mattstep has quit IRC | 07:21 | |
| *** mattstep has joined #openstack-dev | 07:21 | |
| soren | zykes-: pong | 07:31 |
| openstackgerrit | A change was merged to openstack/quantum: Fix flag name for l3 agent external network id https://review.openstack.org/13620 | 07:39 |
| *** eglynn has quit IRC | 07:40 | |
| openstackgerrit | A change was merged to openstack/quantum: clean notification options in quantum.conf. https://review.openstack.org/13695 | 07:46 |
| *** zing has joined #openstack-dev | 07:46 | |
| *** m4xmr has joined #openstack-dev | 07:48 | |
| *** m4xmr has left #openstack-dev | 07:48 | |
| *** adjohn has joined #openstack-dev | 07:52 | |
| *** bencherian has quit IRC | 07:52 | |
| *** termie has quit IRC | 07:55 | |
| *** termie has joined #openstack-dev | 07:55 | |
| *** termie has quit IRC | 07:55 | |
| *** termie has joined #openstack-dev | 07:55 | |
| *** dolphm has quit IRC | 07:56 | |
| *** erikzaadi has joined #openstack-dev | 07:57 | |
| *** danpb has joined #openstack-dev | 07:58 | |
| *** eglynn has joined #openstack-dev | 08:01 | |
| *** darraghb has joined #openstack-dev | 08:07 | |
| *** pixelbeat has joined #openstack-dev | 08:16 | |
| *** sergiutoderascu has joined #openstack-dev | 08:21 | |
| *** zbitter is now known as zaneb | 08:24 | |
| *** almaisan-away is now known as al-maisan | 08:25 | |
| *** sergiutoderascu has quit IRC | 08:27 | |
| *** derekh has joined #openstack-dev | 08:27 | |
| *** sergiutoderascu has joined #openstack-dev | 08:31 | |
| *** \etc\bin has quit IRC | 08:32 | |
| *** sergiutoderascu has quit IRC | 08:32 | |
| *** sergiutoderascu has joined #openstack-dev | 08:32 | |
| *** EmilienM has joined #openstack-dev | 08:39 | |
| *** EmilienM has quit IRC | 08:39 | |
| *** EmilienM has joined #openstack-dev | 08:40 | |
| *** apevec has joined #openstack-dev | 08:40 | |
| *** apevec has joined #openstack-dev | 08:40 | |
| *** gongysh has joined #openstack-dev | 08:48 | |
| gongysh | danwent: hello | 08:49 |
| *** e1mer has joined #openstack-dev | 08:49 | |
| *** e1mer has quit IRC | 08:51 | |
| *** mattstep has quit IRC | 08:54 | |
| *** mattstep has joined #openstack-dev | 08:54 | |
| *** e1mer has joined #openstack-dev | 08:57 | |
| *** adjohn has quit IRC | 08:58 | |
| zykes- | danwent: I tested backporting the Nova floating ip stuff to Folsom from Trunk (The proxy call stuff) and dandy, it works *almost* perfectly https://bugs.launchpad.net/quantum/+bug/1031119 | 09:12 |
| uvirtbot | Launchpad bug 1031119 in quantum "nova: proxy floating ip calls to quantum" [Medium,In progress] | 09:12 |
| zykes- | only thing that doesn't work because of it's own code is horizon | 09:12 |
| amotoki | zykes-: "only thing" is https://bugs.launchpad.net/horizon/+bug/1052561? | 09:14 |
| uvirtbot | Launchpad bug 1052561 in horizon "Needs to accept UUID as ID of Floating IP" [Undecided,In progress] | 09:14 |
| zykes- | amotoki: nice :) | 09:14 |
| zykes- | would be super if distros included this in packaging *pokes adam_g* | 09:15 |
| gongysh | garyk: hi | 09:15 |
| garyk | hi | 09:15 |
| sergiutoderascu | Hello | 09:16 |
| gongysh | about the demo setup in admin guide, are u planning to come up with a new patch? | 09:16 |
| gongysh | Heard that you and dan have a agreement about its change. | 09:17 |
| sergiutoderascu | is there a fix for quantum level3? | 09:17 |
| garyk | gongysh: yes | 09:17 |
| garyk | gongysh: i'll have it ready in about an hour. i did not understand your comment regarding the link. sorry | 09:17 |
| gongysh | garyk: which one. | 09:18 |
| gongysh | ? | 09:18 |
| garyk | https://review.openstack.org/#/c/13569/6/doc/src/docbkx/openstack-network-connectivity-admin/app_demo.xml line 160 | 09:18 |
| gongysh | oh, | 09:20 |
| gongysh | section of core_plugins is Core Plugins, or quantum services | 09:20 |
| gongysh | the title of section of core_plugins is Core Plugins. | 09:21 |
| garyk | ok | 09:24 |
| ttx | vuntz: congrats on your nomination as openSUSE Board Chairman! Somehow I missed that. | 09:27 |
| vuntz | ttx: thanks! | 09:27 |
| * vuntz is having fun trying to find tickets so he can attend both the openstack summit and the opensuse conference | 09:28 | |
| *** dachary has joined #openstack-dev | 09:29 | |
| ttx | SUSE should provide you with your own private jet now | 09:29 |
| *** shang has quit IRC | 09:29 | |
| vuntz | heh | 09:29 |
| dachary | Hi, I sent a request to https://launchpad.net/~openstack-cla/+join about a week ago and it's still waiting to be approved. How long does it typically take ? I've checked my mailbox for additional information request but did not find any. | 09:30 |
| ttx | dachary: I can look into that | 09:30 |
| dachary | ttx: \o | 09:30 |
| *** darjeeling has joined #openstack-dev | 09:31 | |
| dachary | ttx: the email is loic@dachary.org | 09:31 |
| ttx | dachary: can't find you on http://wiki.openstack.org/Contributors ? | 09:32 |
| ttx | dachary: you're supposed to reference your CLA on that wiki page | 09:32 |
| ttx | (see http://wiki.openstack.org/HowToContribute#If_you.27re_a_developer.2C_start_here: ) | 09:32 |
| dachary | ttx: thanks for your patience and my apologies for the noise. My mistake entirely :-( | 09:33 |
| ttx | (and yes, we are working on simplifying that process) | 09:34 |
| *** dachary has quit IRC | 09:43 | |
| *** dachary1 has joined #openstack-dev | 09:43 | |
| *** smooc has joined #openstack-dev | 10:02 | |
| *** kyriakos has joined #openstack-dev | 10:08 | |
| *** MOsyda has joined #openstack-dev | 10:18 | |
| *** mnewby has joined #openstack-dev | 10:21 | |
| *** zhuadl has quit IRC | 10:21 | |
| *** al-maisan is now known as almaisan-away | 10:30 | |
| *** smooc has quit IRC | 10:37 | |
| *** mnewby has quit IRC | 10:45 | |
| dhellmann | eglynn: ping? | 10:45 |
| eglynn | dhellmann: pong | 10:46 |
| dhellmann | responding to your ping from last night | 10:46 |
| eglynn | dhellmann: ah yes, thanks ... I was just wondering why the compute handler doesn't consume resize notifications (to take account of the delta in RAM and/or cores) | 10:47 |
| eglynn | dhellmann: is it because the message payload doesn't include the old instance type? | 10:47 |
| dhellmann | it may just be we haven't gotten to that one yet | 10:47 |
| eglynn | dhellmann: a-ha, ok ... I'll add it my list | 10:47 |
| dhellmann | would you open a bug ticket? | 10:47 |
| eglynn | dhellmann: sure | 10:48 |
| dhellmann | I don't know how resizing works, so I couldn't describe what needs to be done | 10:48 |
| dhellmann | if we still get "exists" events we may not need to worry about the resize | 10:48 |
| *** mnewby has joined #openstack-dev | 10:48 | |
| dhellmann | although it would be good to have the info as soon as it is relevant | 10:48 |
| eglynn | dhellmann: one other slightly related q ... slightly confused as to why instance start and delete are handled equivalently | 10:49 |
| eglynn | dhellmann: I would have thought a delete would be a delta/-1 as opposed to a absolute/1 on the instances counter | 10:49 |
| dhellmann | that has to do with the way we use the meter data in the database | 10:50 |
| dhellmann | we don't track start and end as different types of events | 10:50 |
| dhellmann | rather, we record state at intervals during the lifetime of the resource | 10:50 |
| dhellmann | so when we get the delete event the resource still exists, and record that | 10:50 |
| dhellmann | we don't expect any other events, so that last event will be used as the end of the lifetime when asking questions about the resource | 10:51 |
| dhellmann | but there isn't anything in the record to indicate that it is the last one, other than having the maximum timestamp | 10:51 |
| eglynn | dhellmann: a-ha OK, I see .. hence the compute.instance.delete.start interest as opposed to compute.instance.delete.end | 10:51 |
| eglynn | got it | 10:52 |
| dhellmann | exactly | 10:52 |
| eglynn | thanks, that clears it up | 10:52 |
| dhellmann | don't charge the customer for the time it takes us to shutdown their instance | 10:52 |
| eglynn | dhellmann: I wonder do we need to slightly adjust the volume event_types in the light of that, from volume.delete.end to volume.delete.start | 10:54 |
| dhellmann | ah, yeah, I hadn't noticed that, good call | 10:55 |
| eglynn | dhellmann: cool, I'll take of it ... | 10:56 |
| dhellmann | please open a ticket? we're experiencing a burst of activity right now, so I want to make sure nothing gets lost or forgotten :-) | 10:58 |
| eglynn | dhellmann: will do for both | 10:58 |
| dhellmann | thanks! | 10:58 |
| dhellmann | btw, what timezone are you in? | 10:59 |
| eglynn | dhellman: GMT (Ireland) | 11:00 |
| dhellmann | eglynn: thanks, that was my guess. I'm in US Eastern, so you *just* missed me signing off for the evening last night | 11:01 |
| *** smooc has joined #openstack-dev | 11:01 | |
| eglynn | dhellmann: no worries, thanks for getting back to me today | 11:02 |
| dhellmann | eglynn: any time, thanks for your work on ceilometer! | 11:02 |
| eglynn | np, its been an interesting ramp-up | 11:02 |
| dhellmann | :-) | 11:03 |
| hugokuo | where's .meta file come from in swift ? | 11:05 |
| dhellmann | eglynn: could you join us in #openstack-metering? | 11:11 |
| eglynn | dhellmann: joining ... | 11:12 |
| *** wenjianhn has quit IRC | 11:12 | |
| *** eglynn is now known as eglynn-AFK | 11:19 | |
| zykes- | any cinder people here ? | 11:28 |
| *** romainf has joined #openstack-dev | 11:31 | |
| *** sandywalsh_ has joined #openstack-dev | 11:36 | |
| *** sandywalsh has quit IRC | 11:36 | |
| *** zhuadl has joined #openstack-dev | 11:41 | |
| *** maurosr has joined #openstack-dev | 11:43 | |
| *** sandywalsh_ has quit IRC | 11:46 | |
| davidkranz | jgriffith: I was out yesterday but see there was some discussion of the volume test failures in tempest. Is there any more information? | 11:55 |
| davidkranz | jgriffith: The failures happen randomly on both the cinder and volumes variant of the tempest runs. | 11:55 |
| *** mrunge has quit IRC | 11:56 | |
| *** gongysh has quit IRC | 12:00 | |
| *** dhellmann is now known as dhellmann-afk | 12:03 | |
| *** markvoelker has joined #openstack-dev | 12:05 | |
| *** zz_Kiall is now known as Kiall | 12:05 | |
| *** Tross has quit IRC | 12:11 | |
| *** gongysh has joined #openstack-dev | 12:15 | |
| *** dachary has joined #openstack-dev | 12:15 | |
| gongysh | garyk: where are u? | 12:16 |
| garyk | gongysh: i am lost in docbook :) | 12:16 |
| *** dachary1 has quit IRC | 12:17 | |
| *** alunduil has quit IRC | 12:18 | |
| *** markmc has joined #openstack-dev | 12:19 | |
| *** eglynn-AFK is now known as eglynn | 12:20 | |
| *** tgall_foo has joined #openstack-dev | 12:24 | |
| *** Kiall is now known as zz_Kiall | 12:33 | |
| garyk | gongysh: amotoki: thanks for the reviews. i have posted another patch. would it also be possible that you take a look at https://review.openstack.org/#/c/13743/ | 12:35 |
| zykes- | Hmm, on the L3 code any Quantum people should qg-ab5dea8c-be not be created as a device when it adds a port to br-ex ? | 12:36 |
| zykes- | http://paste.openstack.org/show/21150/ | 12:37 |
| amotoki | garyk: thanks for updating. I have one question on the demo. | 12:39 |
| garyk | amotoki: ok | 12:39 |
| amotoki | garyk: The router is created by 'demo' user. Is it intended? | 12:39 |
| *** zhuadl has quit IRC | 12:39 | |
| garyk | amotoki: i think so - do you think that that this should be admin? | 12:40 |
| *** zhuadl has joined #openstack-dev | 12:40 | |
| amotoki | garyk: both cases work fine. The demo setup is similar to "Provider router with private networks" in the overview section. | 12:41 |
| amotoki | In this case, the router is created by admin. | 12:42 |
| garyk | amotoki: ok. i'll read that section and get back to you | 12:42 |
| gongysh | garyk: the picture does not fit well in PDF. | 12:42 |
| garyk | gongysh: i'll check | 12:42 |
| garyk | gongysh: i run mvn generate-sources - i do not see a pdf file created. what am i missing? | 12:43 |
| amotoki | garyk: thanks. In addition, I commented some typos. Please check them. | 12:44 |
| garyk | amotoki: ok, will do. | 12:44 |
| amotoki | garyk: Don't you have target/docbkx/webhelp/unknown/openstack-network/admin/bk-quantum-admin-guide-trunk.pdf? | 12:45 |
| amotoki | under "openstack-network-connectivity-admin" directory | 12:45 |
| garyk | amotoki: thanks. i see it now | 12:46 |
| garyk | gongysh: it looks good on page 35. is this too big? | 12:49 |
| *** kbringard has joined #openstack-dev | 12:49 | |
| *** dhellmann-afk is now known as dhellmann | 12:50 | |
| *** krtaylor has joined #openstack-dev | 12:51 | |
| *** dhellmann has quit IRC | 12:52 | |
| *** tgall_foo has quit IRC | 12:53 | |
| *** gongysh_ has joined #openstack-dev | 12:53 | |
| gongysh_ | garyk: pdf page 35 | 12:53 |
| garyk | gongysh_: yes, page 35 | 12:54 |
| *** gongysh has quit IRC | 12:54 | |
| amotoki | garyk, gongysh_: It also looks good in my env (with the latest patch set). | 12:54 |
| gongysh_ | garyk, amotoki: I am still at patch 7. I am building patch 9 now. | 12:56 |
| garyk | gongysh_: ok. thanks | 12:56 |
| gongysh_ | garyk: can we specify which user case in the overview part we are describing? | 13:00 |
| garyk | gongysh_: are you talking about the start of appendix a? | 13:03 |
| *** rkukura has quit IRC | 13:05 | |
| gongysh_ | I mean which use case described in overview part by dan the demo setup is trying to configure. | 13:06 |
| garyk | gongysh_: good question. i need to read his overview :) | 13:07 |
| *** dprince has joined #openstack-dev | 13:07 | |
| *** dachary1 has joined #openstack-dev | 13:09 | |
| gongysh_ | garyk: we need to take care of each other when we write good books. :) | 13:09 |
| garyk | gongysh_: true. | 13:09 |
| garyk | gongysh_: i would say that it covers the case "Provider Router with Private Networks" | 13:10 |
| *** ewindisch has joined #openstack-dev | 13:10 | |
| *** ewindisch has quit IRC | 13:11 | |
| *** salv-orlando has joined #openstack-dev | 13:13 | |
| *** dachary has quit IRC | 13:13 | |
| *** dachary1 is now known as dachary | 13:13 | |
| gongysh_ | yes. If you think it is, let us write it down at the beginning of our demo setup. | 13:13 |
| gongysh_ | Dare we? :) | 13:14 |
| salv-orlando | Hey zykes- … sorry for dropping out yesterday.I've looked at the output you've posted. I see the output from br-int and br-virtual. Is br-virtual the one you use for the external gateway? | 13:14 |
| *** romainf has quit IRC | 13:14 | |
| EmilienM | salv-orlando: Hi | 13:15 |
| EmilienM | salv-orlando: I could see last night a patch has been commited into nova for HA queues with RabbitMQ | 13:16 |
| EmilienM | salv-orlando: I was thinking if I could try to create a blueprint & integrate the code for Quantum | 13:16 |
| salv-orlando | EmilienM: That's ok. I hope you don't want to target Folsom as I am afraid it is too late | 13:18 |
| EmilienM | salv-orlando: no I don't | 13:18 |
| salv-orlando | salv-orlando: great. | 13:18 |
| EmilienM | salv-orlando: please let me know if I do something wrong | 13:19 |
| salv-orlando | I'm self addressing myself on the IRC… I need a doctor :) | 13:19 |
| EmilienM | salv-orlando: that'll be actually my first bp and so on | 13:19 |
| EmilienM | salv-orlando: /o\ | 13:19 |
| salv-orlando | EmilienM: there's nothing wrong in creating a blueprint. Target Grizzly series (I don't think milestone are already in place). | 13:19 |
| EmilienM | ok | 13:19 |
| salv-orlando | And if the blueprint needs some explanations concerning design or implementation, please provide all the necessary details. | 13:20 |
| EmilienM | I'll try | 13:20 |
| EmilienM | salv-orlando: congrats for the doc, it's more clear since the last days | 13:20 |
| *** spn has quit IRC | 13:21 | |
| *** spn has joined #openstack-dev | 13:22 | |
| *** zhuadl has quit IRC | 13:23 | |
| *** sandywalsh has joined #openstack-dev | 13:30 | |
| *** sandywalsh has quit IRC | 13:30 | |
| *** mrodden has joined #openstack-dev | 13:30 | |
| *** tgall_foo has joined #openstack-dev | 13:31 | |
| *** tgall_foo has quit IRC | 13:31 | |
| *** tgall_foo has joined #openstack-dev | 13:31 | |
| *** roge has joined #openstack-dev | 13:31 | |
| EmilienM | salv-orlando: Need I specify an approver or a drafter ? | 13:34 |
| gongysh_ | garyk: done? | 13:34 |
| *** tgall_foo has quit IRC | 13:34 | |
| garyk | gongysh_: in a minute. i am just adding in a link to the section... | 13:35 |
| gongysh_ | garyk: fine, waiting ... | 13:35 |
| *** e1mer has quit IRC | 13:36 | |
| salv-orlando | EmilienM: the PTL (danwent) should be approver for all bps | 13:37 |
| *** tgall_foo has joined #openstack-dev | 13:37 | |
| *** tgall_foo has quit IRC | 13:37 | |
| *** tgall_foo has joined #openstack-dev | 13:37 | |
| EmilienM | salv-orlando: great | 13:37 |
| EmilienM | https://blueprints.launchpad.net/quantum/+spec/high-available-quantum-queues-in-rabbitmq | 13:37 |
| EmilienM | salv-orlando: ^ | 13:37 |
| salv-orlando | EmilienM: thanks | 13:37 |
| EmilienM | salv-orlando: can I start to code something ? or... | 13:39 |
| salv-orlando | EmilienM: grizzly gates are open | 13:39 |
| *** alunduil has joined #openstack-dev | 13:40 | |
| EmilienM | ok | 13:40 |
| garyk | gongysh_: i am done (hopefully) | 13:40 |
| *** alunduil has quit IRC | 13:42 | |
| *** alunduil has joined #openstack-dev | 13:43 | |
| MOsyda | Hi. I've installed (using devstack) nova folsom-rc3 and I have strange problem... Heat managed to create stack, instances got their ips, so whole 'system' worked well | 13:47 |
| MOsyda | ...but I cant do 'nova list' - there is strange ip in debug: 10.0.2.15 Output: http://pastie.org/4829667 | 13:47 |
| MOsyda | where it might come from? | 13:48 |
| *** mtreinish has joined #openstack-dev | 13:51 | |
| zz_Kiall | Anyone know if Patrick Galbraith from HP hangs around these parts? | 13:51 |
| *** zz_Kiall is now known as Kiall | 13:52 | |
| gongysh_ | garyk: another patch is needed. | 13:53 |
| garyk | gongysh_: ok | 13:55 |
| garyk | in 2 minutes | 13:55 |
| garyk | gongysh_: done :) | 13:59 |
| garyk | gongysh_: sorry, it took 4 minutes | 13:59 |
| *** aeperezt has joined #openstack-dev | 14:02 | |
| *** tgall_foo has quit IRC | 14:05 | |
| *** dachary has quit IRC | 14:05 | |
| *** mrodden has quit IRC | 14:06 | |
| *** japage has joined #openstack-dev | 14:07 | |
| *** maoy has joined #openstack-dev | 14:10 | |
| *** littleidea has joined #openstack-dev | 14:11 | |
| *** ghe has joined #openstack-dev | 14:15 | |
| *** ghe is now known as GheRivero | 14:15 | |
| *** spn has quit IRC | 14:16 | |
| *** spn has joined #openstack-dev | 14:16 | |
| *** metral has joined #openstack-dev | 14:16 | |
| *** spn is now known as spn_ | 14:17 | |
| *** tgall_foo has joined #openstack-dev | 14:19 | |
| EmilienM | salv-orlando: https://review.openstack.org/#/c/13760/ | 14:21 |
| *** gongysh_ has quit IRC | 14:22 | |
| *** mrodden has joined #openstack-dev | 14:26 | |
| *** belliott has quit IRC | 14:29 | |
| *** belliott has joined #openstack-dev | 14:30 | |
| *** GheRivero has quit IRC | 14:31 | |
| *** Gordonz has joined #openstack-dev | 14:31 | |
| *** salv-orlando has quit IRC | 14:32 | |
| *** Gordonz has quit IRC | 14:33 | |
| *** Gordonz has joined #openstack-dev | 14:34 | |
| *** dhellmann has joined #openstack-dev | 14:35 | |
| *** gongysh_ has joined #openstack-dev | 14:36 | |
| *** torandu has quit IRC | 14:41 | |
| *** Mandell has quit IRC | 14:42 | |
| *** spn_ has quit IRC | 14:43 | |
| *** spn has joined #openstack-dev | 14:44 | |
| *** GheRivero has joined #openstack-dev | 14:45 | |
| *** rnirmal has joined #openstack-dev | 14:45 | |
| *** dolphm has joined #openstack-dev | 14:46 | |
| *** andrewbogott_afk is now known as andrewbogott | 14:48 | |
| *** GheRivero is now known as Ghe_Rivero | 14:51 | |
| *** rkukura has joined #openstack-dev | 14:51 | |
| *** rnirmal has quit IRC | 14:53 | |
| *** japage has quit IRC | 14:54 | |
| *** japage has joined #openstack-dev | 14:56 | |
| *** jtran has joined #openstack-dev | 14:56 | |
| *** japage has left #openstack-dev | 14:58 | |
| *** datsun180b has joined #openstack-dev | 14:58 | |
| *** MOsyda has quit IRC | 15:01 | |
| *** journeeman has joined #openstack-dev | 15:02 | |
| *** torandu has joined #openstack-dev | 15:04 | |
| *** Ghe_Rivero has quit IRC | 15:05 | |
| *** mrodden has quit IRC | 15:05 | |
| *** adjohn has joined #openstack-dev | 15:07 | |
| *** jaypipes has joined #openstack-dev | 15:08 | |
| *** Tross has joined #openstack-dev | 15:08 | |
| *** cp16net is now known as cp16net|away | 15:09 | |
| *** mjfork has quit IRC | 15:09 | |
| *** cp16net|away is now known as cp16net | 15:09 | |
| *** Tross has left #openstack-dev | 15:10 | |
| *** mjfork has joined #openstack-dev | 15:10 | |
| *** dachary has joined #openstack-dev | 15:10 | |
| *** Yar has joined #openstack-dev | 15:13 | |
| ttx | Bugmail spam provided courtesy of Folsom release process. | 15:13 |
| *** reidrac has quit IRC | 15:13 | |
| *** mrodden has joined #openstack-dev | 15:13 | |
| *** kbringard has quit IRC | 15:13 | |
| *** maoy has quit IRC | 15:15 | |
| *** smooc has quit IRC | 15:16 | |
| zykes- | ttx: get launchpad to disable that :p | 15:17 |
| ttx | zykes-: rest assured, I disabled it for me | 15:17 |
| jgriffith | ttx: I like it, my boss stopped by and thinks I must be *important* based on the massive number of emails | 15:18 |
| zykes- | btw ttx if someone needs to "backport" floating support in holsom for nova + quantum and horizon: http://cloudistic.me/ | 15:18 |
| jgriffith | ttx: Either that or that guy Thierry is just a PITA | 15:18 |
| ttx | "Those French people sure are chatty" | 15:19 |
| Yar | danwent: Hello, here is paste for my bug http://paste.openstack.org/show/21153/ | 15:19 |
| *** jemartin has joined #openstack-dev | 15:21 | |
| zul | jgriffith: but you arent | 15:21 |
| jgriffith | zul: alas... I am not :( | 15:21 |
| zykes- | gongysh_: ping | 15:21 |
| zul | jgriffith: join the club :) | 15:22 |
| jgriffith | zul: LOL | 15:22 |
| Yar | danwent: and you was right - router was created by demo. So the problem rises only when admin add interface to demo subnet | 15:22 |
| *** dragondm has left #openstack-dev | 15:26 | |
| *** mindpixel has quit IRC | 15:26 | |
| *** gongysh__ has joined #openstack-dev | 15:28 | |
| hugokuo | does anyone knows the usag of .meta in swift ? | 15:29 |
| danwent | Yar: hi | 15:29 |
| *** markmcclain has joined #openstack-dev | 15:31 | |
| *** Tross has joined #openstack-dev | 15:31 | |
| *** Tross has left #openstack-dev | 15:31 | |
| danwent | Yar: yes, the router is owned by demo, not admin (perhaps I misunderstood your original comment on the bug) | 15:31 |
| *** gongysh_ has quit IRC | 15:31 | |
| danwent | btw, the two main deployment modes are listed in the admin guideE: http://docs.openstack.org/trunk/openstack-network/admin/content/use_cases_single_router.html | 15:31 |
| danwent | http://docs.openstack.org/trunk/openstack-network/admin/content/use_cases_tenant_router.html | 15:32 |
| Yar | danwent: oh, I've missed this document, thanks! | 15:33 |
| danwent | Yar: thanks for the testing. will continue discussion in our other thread | 15:34 |
| *** markmcclain has quit IRC | 15:35 | |
| *** rnirmal has joined #openstack-dev | 15:37 | |
| *** ghe has joined #openstack-dev | 15:40 | |
| *** ghe is now known as Guest45846 | 15:40 | |
| *** lloydde has joined #openstack-dev | 15:41 | |
| *** Guest45846 is now known as Ghe_Rivero | 15:41 | |
| openstackgerrit | A change was merged to openstack/nova: Check that an image is active before spawning instances. https://review.openstack.org/13692 | 15:43 |
| *** salv-orlando has joined #openstack-dev | 15:43 | |
| *** journeeman has quit IRC | 15:43 | |
| *** littleidea has quit IRC | 15:45 | |
| *** dragondm has joined #openstack-dev | 15:45 | |
| *** Ghe_Rivero has quit IRC | 15:47 | |
| *** kbringard has joined #openstack-dev | 15:49 | |
| *** dhellmann has quit IRC | 15:50 | |
| *** srini_g has joined #openstack-dev | 15:51 | |
| *** bencherian has joined #openstack-dev | 15:55 | |
| *** rafaduran has quit IRC | 15:57 | |
| *** reed has joined #openstack-dev | 15:59 | |
| *** erikzaadi has quit IRC | 16:00 | |
| *** littleidea has joined #openstack-dev | 16:01 | |
| *** markmc has quit IRC | 16:02 | |
| *** reed_ has joined #openstack-dev | 16:02 | |
| *** reed has quit IRC | 16:03 | |
| *** dachary has quit IRC | 16:05 | |
| *** hemna has joined #openstack-dev | 16:05 | |
| *** dachary has joined #openstack-dev | 16:05 | |
| *** reed_ is now known as reed | 16:07 | |
| ttx | It's out. | 16:07 |
| garyk | n0ano: ping | 16:07 |
| n0ano | garyk, ack | 16:07 |
| ttx | OpenStack 2012.2 ("Folsom") is out, great job everyone | 16:07 |
| garyk | n0ano: you around to talk | 16:07 |
| reed | Long live Folsom | 16:08 |
| n0ano | sure, what's up? | 16:08 |
| reed | at least 6 months at least | 16:08 |
| *** darraghb has quit IRC | 16:08 | |
| garyk | n0ano: nachi? | 16:08 |
| n0ano | garyk, nachi - sorry, I don't understand | 16:08 |
| garyk | n0ano: sorry wrong person | 16:08 |
| n0ano | garyk, and I have such a unique handle :-) NP. | 16:09 |
| *** adalbas has quit IRC | 16:09 | |
| ttx | reed: now if only Launchpad wouldn't timeout serving the Nova download page... | 16:09 |
| reed | ehehe | 16:09 |
| *** andrewbogott is now known as andrewbogott_afk | 16:11 | |
| *** Ghe_Rivero has joined #openstack-dev | 16:12 | |
| *** zaitcev has joined #openstack-dev | 16:12 | |
| zykes- | ttx: is there docs for Cinder anywhere ? | 16:12 |
| zykes- | I can't find them | 16:13 |
| ttx | zykes-: jgriffith should be able to help you out | 16:13 |
| * zykes- sends a ping to jgriffith | 16:13 | |
| *** epim has quit IRC | 16:14 | |
| reed | ttx, where are the download pages? | 16:14 |
| *** sandywalsh has joined #openstack-dev | 16:15 | |
| ttx | that's the launchpad milestone pages that are referenced in the announcement | 16:15 |
| *** Mandell has joined #openstack-dev | 16:15 | |
| ttx | reed: ^ | 16:15 |
| reed | got it, thanks | 16:15 |
| ttx | reed: if you urgently need the nova tarball, it's actually also available at https://launchpad.net/nova/folsom/folsom-rc3 < it's the same tarball | 16:16 |
| *** Ghe_Rivero has quit IRC | 16:18 | |
| *** eglynn has quit IRC | 16:20 | |
| *** adalbas has joined #openstack-dev | 16:21 | |
| *** dripton has joined #openstack-dev | 16:21 | |
| zykes- | danwent: or salv-orlando one of you guys around ? pref salv-orlando atm ;) | 16:22 |
| apevec | ttx, https://launchpad.net/glance/ shows "Latest version is 2012.1.2 " | 16:22 |
| apevec | but folsom tarball is there, https://launchpad.net/glance/+download?direction=backwards&memo=30&start=20 | 16:23 |
| ttx | apevec: need to swap series | 16:23 |
| ttx | just a sec | 16:23 |
| *** eharney has joined #openstack-dev | 16:23 | |
| *** mnewby_ has joined #openstack-dev | 16:24 | |
| *** mlavalle has joined #openstack-dev | 16:26 | |
| *** mnewby has quit IRC | 16:26 | |
| *** mnewby_ is now known as mnewby | 16:26 | |
| *** nunosantos has joined #openstack-dev | 16:28 | |
| *** ev0ldave has joined #openstack-dev | 16:28 | |
| reed | annegentle, any ETA for the install guides for Folsom? | 16:29 |
| salv-orlando | I am in a meeting - will be with you soon | 16:30 |
| salv-orlando | zykes-: ^ ^ | 16:31 |
| zykes- | ok :) | 16:31 |
| *** nunosantos_ has joined #openstack-dev | 16:31 | |
| *** darraghb has joined #openstack-dev | 16:31 | |
| zykes- | salv-orlando: remember yesterdays problem with missing interfaces for l3 ? | 16:31 |
| salv-orlando | yes | 16:31 |
| zykes- | What can be the error ? | 16:32 |
| *** koolhead17 has joined #openstack-dev | 16:32 | |
| *** armaan has joined #openstack-dev | 16:35 | |
| *** titankiller has joined #openstack-dev | 16:36 | |
| ttx | apevec: fixed for glance, now I have to fix it for others | 16:37 |
| apevec | others were good when I checked earlier | 16:37 |
| apevec | ok, nova is not | 16:37 |
| ttx | apevec: it's a bit more complicated than it seems | 16:38 |
| apevec | ttx, glance is good now, thanks | 16:38 |
| ttx | once you switch current development to grizzly... folsom becomes yet another supported release | 16:38 |
| ttx | and somehow LP picks essex as "latest" | 16:38 |
| *** eglynn has joined #openstack-dev | 16:38 | |
| *** soosiechoi has joined #openstack-dev | 16:39 | |
| ttx | Due to https://bugs.launchpad.net/launchpad/+bug/980749 | 16:39 |
| uvirtbot | Launchpad bug 980749 in launchpad ""Downloads" portlet shows *older* version than current stable" [Low,Triaged] | 16:39 |
| ttx | so you have to say that essex is "obsolete" so that LP knows which one to display. | 16:40 |
| *** ghe has joined #openstack-dev | 16:40 | |
| *** ghe is now known as Guest75383 | 16:41 | |
| * ttx is working on getting a bigger timeout on that Nova LP download page | 16:41 | |
| *** Guest75383 is now known as Ghe_Rivero | 16:43 | |
| *** markmcclain has joined #openstack-dev | 16:44 | |
| *** danpb has quit IRC | 16:45 | |
| dachary | I'm now on the http://wiki.openstack.org/Contributors page ttx. How long does it typically takes to process an application ? I'll set a reminder and I'm not in a hurry ;-) | 16:46 |
| ttx | dachary: will look into it now | 16:46 |
| mordred | davidkranz: ping | 16:47 |
| openstackgerrit | A change was merged to openstack/nova: Modified 404 error response to show specific message https://review.openstack.org/13694 | 16:47 |
| dachary | ttx: thanks ! | 16:48 |
| ttx | dachary: done | 16:48 |
| * dachary hugs ttx | 16:49 | |
| zykes- | how is HA in openstack ? | 16:49 |
| *** zbitter has joined #openstack-dev | 16:49 | |
| zykes- | for VMs | 16:50 |
| *** apevec has quit IRC | 16:50 | |
| *** zaitcev has quit IRC | 16:51 | |
| reed | zykes-, it's a highly debated topic... some google results http://www.mirantis.com/blog/ha-platform-components-mysql-rabbitmq/ http://www.sebastien-han.fr/blog/2012/06/28/openstack-glance-keystone-ha/ | 16:51 |
| *** zaneb has quit IRC | 16:52 | |
| reed | http://prezi.com/5pg6wdmmxoc1/high-availability-in-openstack/ | 16:52 |
| *** garyk has quit IRC | 16:52 | |
| *** jog0 has joined #openstack-dev | 16:53 | |
| *** derekh has quit IRC | 16:53 | |
| zykes- | reed: that's for mysql etc | 16:53 |
| zykes- | not for VMs | 16:53 |
| zykes- | is host aggregates still in or are they getting removed? | 16:53 |
| *** dhellmann has joined #openstack-dev | 16:54 | |
| *** m4xmr has joined #openstack-dev | 16:54 | |
| *** armaan has left #openstack-dev | 16:55 | |
| davidkranz | mordred: Here now. | 16:55 |
| *** JStoker has quit IRC | 16:57 | |
| *** m4xmr has quit IRC | 16:58 | |
| mordred | davidkranz: was pinging about a further info request you made on a summit talk thing ... but then I realized I can put notes in the proposal :) | 16:58 |
| *** JStoker has joined #openstack-dev | 16:58 | |
| davidkranz | mordred: OK, I'll look there. | 16:59 |
| *** zykes- has quit IRC | 17:00 | |
| davidkranz | mordred: That's cool. If we're lucky we could kill two birds with one stone here. | 17:00 |
| dragondm | anyone have time for a review? https://review.openstack.org/#/c/13594/ | 17:00 |
| *** zykes- has joined #openstack-dev | 17:01 | |
| *** anuj has joined #openstack-dev | 17:02 | |
| *** zing has quit IRC | 17:02 | |
| jog0 | zykes-: still in | 17:04 |
| *** JStoker has quit IRC | 17:05 | |
| *** dhellmann has quit IRC | 17:05 | |
| zykes- | jog0: yes | 17:06 |
| *** dhellmann has joined #openstack-dev | 17:06 | |
| mordred | davidkranz: yes, I believe we can... I think there's actually three birds in there... | 17:09 |
| *** jdurgin has joined #openstack-dev | 17:09 | |
| zykes- | jog0: ? | 17:10 |
| *** JStoker has joined #openstack-dev | 17:12 | |
| *** kyriakos has quit IRC | 17:12 | |
| *** ayoung has joined #openstack-dev | 17:13 | |
| *** danwent has quit IRC | 17:14 | |
| *** eglynn has quit IRC | 17:15 | |
| jgriffith | davidkranz: ping | 17:17 |
| zykes- | jgriffith: there you are ! :p | 17:17 |
| zykes- | jgriffith: where's the doc for cinder ? | 17:17 |
| * jgriffith runs to hide | 17:17 | |
| zykes- | If I may :L) | 17:17 |
| jgriffith | zykes-: Working on docs.. right now all I have to offer is release notes | 17:17 |
| zykes- | :| | 17:18 |
| *** markmcclain has quit IRC | 17:18 | |
| zykes- | jgriffith: can you help me along to get stuff working ? | 17:18 |
| *** markmcclain has joined #openstack-dev | 17:18 | |
| jgriffith | zykes-: sure, what are you having trouble with? | 17:19 |
| jgriffith | zykes-: FYI, you really should be able to s/nova-vol/cinder/ from current docs for the most part | 17:19 |
| *** soosiechoi has quit IRC | 17:19 | |
| zykes- | jgriffith: basically I get a error saying that I don't have any valid hosts | 17:19 |
| zykes- | even though the cinder-volume service is running and connected to the mq | 17:20 |
| jog0 | zykes-: wait are you talking about cinder? | 17:20 |
| *** PotHix has joined #openstack-dev | 17:20 | |
| zykes- | jog0: both nova and cinder jog0 ;) | 17:21 |
| zykes- | wondering about HA or "VM HA" for nov | 17:21 |
| zykes- | a also | 17:21 |
| zykes- | jog0: if I do cinder-manage host list | 17:22 |
| zykes- | ehm, jgriffith shouldn't it then list out a cnder-volume host as well ? | 17:22 |
| jgriffith | yes | 17:22 |
| jgriffith | zykes-: what are you getting? | 17:23 |
| jog0 | zykes-: yes to nova, no to cinder | 17:23 |
| jog0 | zykes-: for aggregates | 17:23 |
| zykes- | jgriffith: well it doesn | 17:24 |
| jgriffith | zykes-: well ok | 17:24 |
| zykes- | any way I can debug why ? | 17:25 |
| jgriffith | zykes-: Looking | 17:26 |
| jgriffith | zykes-: Check your endpoints form keystone | 17:26 |
| zykes- | jgriffith: eh, isn't cinder services located in the database ? | 17:26 |
| mordred | wow. did we really not get run_tests.py removed from keystone this cycle | 17:26 |
| * mordred is ashamed | 17:26 | |
| mordred | ayoung: my last email to you assumed that I'd made a fix to keystone that I'd made in nova ... so it might not make a lot of sense | 17:27 |
| *** soosiechoi has joined #openstack-dev | 17:28 | |
| jgriffith | zykes-: ? | 17:28 |
| jgriffith | zykes-: I'm trying to figure out if you even have things configured to talk to cinder | 17:28 |
| jgriffith | zykes-: ie endpoints, 'keystone services-list' etc | 17:28 |
| *** anuj has quit IRC | 17:28 | |
| zykes- | ah, jgriffith it needs mysql set in cinder.conf for the volume service as well | 17:28 |
| zykes- | will that be needed in the future or ? | 17:28 |
| jgriffith | zykes-: yes | 17:29 |
| zykes- | jgriffith: why, I mean stuff is sent via RPC no ? | 17:30 |
| *** bencherian has quit IRC | 17:30 | |
| ayoung | mordred, I'm catching up with email right now, and thrashing on concepts... | 17:30 |
| *** rohitk has joined #openstack-dev | 17:30 | |
| *** ewindisch has joined #openstack-dev | 17:30 | |
| ayoung | mordred, what topic was that on again? | 17:31 |
| *** rohitk has quit IRC | 17:31 | |
| mordred | ayoung: monkeypatching in test runner | 17:31 |
| ayoung | ah | 17:31 |
| mordred | ayoung: I just followed up to myself | 17:31 |
| mordred | ayoung: there is no customer test runner any more in nova or glance | 17:31 |
| * mordred has been working to get all of that deleted for a while | 17:31 | |
| ayoung | mordred, in Keystone | 17:31 |
| mordred | s/customer/custom/ | 17:31 |
| mordred | yes. I didn't realize I didn't get to keystone | 17:32 |
| ayoung | I'll send you the commit, one sec | 17:32 |
| mordred | nope, I'm with you - I just made a request in a manner that didn't immediately make sense | 17:32 |
| zykes- | jog0: about aggregates, can they be used for VM HA? | 17:32 |
| ayoung | mordred, so I want a patch like 81cb2ab2 for any test runner | 17:32 |
| jgriffith | zykes-: manager to other cinder nodes yes, but db calls no | 17:32 |
| jgriffith | zykes-: Works the same as nova | 17:32 |
| *** rohitk has joined #openstack-dev | 17:32 | |
| mordred | ayoung: indeed | 17:32 |
| mordred | ayoung: the thing is, we use non-modified nose for nova and glance... but the monkeypatching occurs in a test baseclass/fixture setup | 17:33 |
| jgriffith | make -xtest | 17:33 |
| jgriffith | grrrr | 17:33 |
| jog0 | zykes-: I am not sure what you mean, can you elaborate? | 17:33 |
| ayoung | mordred, not for us it doesn't. And it shouldn't | 17:33 |
| mordred | ayoung: so we can put the switch to disable it the base class | 17:33 |
| mordred | ayoung: it should | 17:33 |
| mordred | ayoung: you should not need to patch the test runner to set up test env | 17:33 |
| *** s0mik has joined #openstack-dev | 17:34 | |
| mordred | because that limits you to just using your modified test runner | 17:34 |
| ayoung | mordred, OK, so long as the test runner allows the switch to be passed on the command line I am all for it | 17:34 |
| ayoung | I'd like to get rid of Eventlet all together, but that is another rant. | 17:34 |
| mordred | well, sure :) | 17:34 |
| zykes- | jog0: as in if one host goes boom will another pick the instances up ? | 17:35 |
| ayoung | So I don't want the tests to be Eventlet specific either | 17:35 |
| mordred | I'll send you a straw-man patch that does what I think you want | 17:35 |
| mordred | ayoung: TOTALLY | 17:35 |
| ayoung | mordred, sounds good to me | 17:35 |
| mordred | ayoung: I think we might ultimately want the same thing, which is for stuff to not be required to be weird | 17:35 |
| mordred | but from different specific directions | 17:35 |
| ayoung | mordred, now we have to talk in Valley Speak for the rest of the day | 17:36 |
| jog0 | zykes-: aggregates are a way to assign metadata to nova-compute nodes. Not to VMs. I am not sure how an aggregate would work for VM HA | 17:36 |
| ayoung | I am so sure. | 17:36 |
| mordred | ayoung: oh. my. GOD! | 17:36 |
| ayoung | mordred, I think we are on the same frequency here | 17:36 |
| *** EmilienM has quit IRC | 17:36 | |
| jeblair | ugh. gag me with a spoon! | 17:37 |
| ayoung | mordred, yeah, I want to use HTTPD, and not have to do custom ports for SELinux reasons. I am with you on "Don't make it weird." | 17:37 |
| ayoung | As if | 17:37 |
| zykes- | Noones thought of VM ha ? | 17:37 |
| mordred | ayoung: so, we moved the eventlet monkey patch in nova to nova/nova/tests/__init__.py - for the disable patch, I can wrap it there... | 17:38 |
| mordred | ayoung: I will need to think a little further on how to do the same thing to make us both happy in keystone | 17:39 |
| mordred | but I'll get it | 17:39 |
| *** brianr has quit IRC | 17:39 | |
| ayoung | mordred, that should be early enough for Keystone. It has to happen before the modules get referenced, as I recall | 17:39 |
| *** brianr has joined #openstack-dev | 17:39 | |
| ayoung | So you want it pretty early on | 17:39 |
| mordred | yes. that's totally correct | 17:39 |
| openstackgerrit | A change was merged to openstack/swift: replace a value set with utils.TRUE_VALUES https://review.openstack.org/12773 | 17:40 |
| zykes- | ayoung: is there a reason to who nova volumes needs to dd the whole volume before deletion ? | 17:40 |
| jog0 | zykes-: AFAIK Heat is trying to do VM HA | 17:40 |
| mordred | ayoung: so, when you said that you don't want the tests to be eventlet specific ... does putting the monkeypatch there for keystone subvert that goal? | 17:40 |
| *** soosiechoi has quit IRC | 17:41 | |
| zykes- | jog0: uhm, ok | 17:41 |
| ayoung | mordred, no, I think that when we can make it optional, we will. Leave it in test setup for now should be fine | 17:41 |
| mordred | kk. awesome | 17:41 |
| zykes- | but jog0 isn't that a nova job ? | 17:41 |
| * mordred goes to hack up a patch... | 17:41 | |
| *** danwent has joined #openstack-dev | 17:42 | |
| zykes- | danwent: ello! | 17:42 |
| danwent | zykes-: yes | 17:42 |
| jog0 | zykes-: That is not clear. amazon treats VMs as ephemeral but rax does not | 17:43 |
| *** darraghb has quit IRC | 17:44 | |
| *** gugl has joined #openstack-dev | 17:44 | |
| zykes- | danwent: wondering if tere | 17:44 |
| *** dansmith has quit IRC | 17:44 | |
| zykes- | if you can help me look at some l3 agent probs, salv-orlando seems to have disappeared | 17:44 |
| *** AlanClark has joined #openstack-dev | 17:45 | |
| danwent | zykes-: have you file a bug on it? I didn't see anything | 17:45 |
| danwent | or a LP question? | 17:45 |
| zykes- | not yet :P | 17:46 |
| salv-orlando | salv-orlando: sorry for disappearing, was in a meeting | 17:47 |
| salv-orlando | I should stop referring to myself | 17:47 |
| salv-orlando | I probably need a doctor | 17:47 |
| zykes- | hah :p | 17:47 |
| zykes- | http://paste.ubuntu.com/1230835/ is basically the problem | 17:47 |
| salv-orlando | zykes-: I can't help thinking if we nail this down we have bug 1051679 | 17:48 |
| uvirtbot | Launchpad bug 1051679 in quantum "Error running ip netns command with l3-agent" [Medium,Incomplete] https://launchpad.net/bugs/1051679 | 17:48 |
| salv-orlando | yesterday you sent me the output of ovs-dpctl | 17:48 |
| *** dansmith has joined #openstack-dev | 17:48 | |
| salv-orlando | but I could not see the bridge for external networking, was it br-virtual? | 17:48 |
| *** rohitk has quit IRC | 17:48 | |
| zykes- | br-ex qg- interfaces are the external part no ? | 17:49 |
| *** rohitk has joined #openstack-dev | 17:49 | |
| salv-orlando | you pasted two links - but I could not see br-ex in none of those | 17:49 |
| *** cp16net is now known as cp16net|away | 17:49 | |
| zykes- | I did | 17:50 |
| zykes- | ovs-dpctl show | 17:50 |
| zykes- | and ovs-vsctl show | 17:50 |
| zykes- | wasn't it that ? | 17:50 |
| salv-orlando | yes, but I could see only br-int and br-virtual | 17:50 |
| mordred | ayoung: sake of argument: https://review.openstack.org/13771 | 17:50 |
| zykes- | http://paste.ubuntu.com/1230847/ < salv-orlando | 17:51 |
| *** andrewbogott_afk is now known as andrewbogott | 17:52 | |
| salv-orlando | if you do at the command line | 17:52 |
| salv-orlando | sudo ip link show qg-ab5dea8c-be you should get an error | 17:52 |
| zykes- | it doens't exist correct | 17:54 |
| *** hemna has quit IRC | 17:54 | |
| *** thingee_zz is now known as thingee | 17:55 | |
| *** hemna has joined #openstack-dev | 17:55 | |
| salv-orlando | zykes-: try the following two commands, in sequence: | 17:56 |
| salv-orlando | sudo ovs-vsctl -- set Interface qg-ab5dea8c-be external-ids:iface-status=inactive | 17:56 |
| salv-orlando | sudo ovs-vsctl -- set Interface qg-ab5dea8c-be external-ids:iface-status=active | 17:56 |
| salv-orlando | I know they look trivial and meaningless, but I need to confirm something that might explain the problem | 17:56 |
| zykes- | sure salv-orlando whatever I can do to make it work! | 17:57 |
| zykes- | didn't help | 17:57 |
| *** Tross has joined #openstack-dev | 17:58 | |
| salv-orlando | so ip link show still tells you the interface does not exist? | 17:58 |
| zykes- | yes | 17:58 |
| salv-orlando | interesting. I guess if you create a different gateway interface you'll still have the same problem? | 17:59 |
| ayoung | mordred, so, the one thing is that for now tests need to skip only the threading monkey patch. You are looking a little further down the road, I think | 17:59 |
| ayoung | SKIP_MONKEYPATCH should be | 17:59 |
| ayoung | standard-threading I think | 17:59 |
| ayoung | but other than that, I think you are on track. I'll give it a test run in a few | 18:00 |
| zykes- | salv-orlando: how you mean, if I make a new subnet and a new gw interface on the router? | 18:00 |
| *** Ghe_Rivero is now known as GheRivero | 18:00 | |
| *** epim has joined #openstack-dev | 18:00 | |
| ayoung | mordred, also, it needs to be passed through from the Python test runner in order to be called from PyDev | 18:01 |
| *** bencherian has joined #openstack-dev | 18:01 | |
| ayoung | That is one reason it was implemented in the python test runner in the first place | 18:01 |
| *** cp16net|away is now known as cp16net | 18:01 | |
| salv-orlando | zykes-: yep | 18:03 |
| salv-orlando | It seems for some reason ovs is not bringing up the interface corresponding to that port | 18:04 |
| mordred | ayoung: what is pydev? | 18:05 |
| ayoung | mordred Eclipse Python IDE. Replace that with debugger and you have the real problem | 18:05 |
| mordred | ayoung: ah! | 18:05 |
| ayoung | mordred, I run the tests from python, not useing the shell script wrapper | 18:05 |
| mordred | ayoung: gotit | 18:05 |
| zykes- | what can the solution be salv-orlando ? | 18:06 |
| *** _adjohn has joined #openstack-dev | 18:06 | |
| *** eharney has quit IRC | 18:06 | |
| mordred | ayoung: (I don't use the shell script wrapper either in most places... but I now grok the thing you are saying... will update) | 18:06 |
| salv-orlando | zykes-: one solution we just tried and did not work :( | 18:06 |
| *** martine has joined #openstack-dev | 18:06 | |
| mordred | ayoung: if we delete run_tests.py and have things just expecting to use nose, can pydev do that? | 18:06 |
| ayoung | rock on, mordred, | 18:06 |
| ayoung | mordred, probably | 18:06 |
| mordred | cool | 18:06 |
| * mordred will investigate | 18:06 | |
| *** adjohn has quit IRC | 18:07 | |
| *** _adjohn is now known as adjohn | 18:07 | |
| salv-orlando | zykes-: did you had this error on every machine where you installed the l3 agent? is it persistent across reboots? | 18:07 |
| *** dspano has joined #openstack-dev | 18:09 | |
| zykes- | I only have 1 l3 server salv-orlando | 18:09 |
| salv-orlando | zykes-: I guess that you've already rebooted it several times :) | 18:10 |
| zykes- | uhuh | 18:10 |
| zykes- | :p | 18:10 |
| salv-orlando | is "uhuh" a "yes, I've rebooted it an awful lot of times"? | 18:11 |
| zykes- | correct :) | 18:11 |
| zykes- | i'm trying to create another network with subnet now and adding a interface to the router with it | 18:11 |
| salv-orlando | ok - so can you try to man ally add another internal interface to br-ex and see if it shows up with ip link show | 18:12 |
| *** rohitk has quit IRC | 18:12 | |
| salv-orlando | I need to exclude all possible non-Quantum related problems | 18:12 |
| zykes- | sudo ovs-vsctl -- --may-exist add-port br-ex qg-test < for example ? | 18:13 |
| *** ewindisch has quit IRC | 18:13 | |
| salv-orlando | and also the set-interface stuff | 18:14 |
| salv-orlando | at least the bit when ypou specify the internal type | 18:14 |
| openstackgerrit | A change was merged to openstack/nova: Loosen anyjson dependency to avoid clash with ceilometer https://review.openstack.org/13744 | 18:14 |
| zykes- | salv-orlando: I ran sudo ovs-vsctl -- --may-exist add-port br-ex qg-test -- set Interface qg-test type=internal -- set Interface qg-test external-ids:iface-id=test -- set Interface qg-test external-ids:iface-status=active -- set Interface qg-test external-ids:attached-mac=fa:16:3e:8b:45:ee < no interface up and running and nothing showing in dmesg either (which it normally does for new interface) | 18:16 |
| salv-orlando | cool. | 18:17 |
| salv-orlando | ovs version? | 18:17 |
| *** garyk has joined #openstack-dev | 18:17 | |
| zykes- | 1.4.0 | 18:17 |
| zykes- | latest in Precise | 18:17 |
| zykes- | Do you want a bug salv-orlando ? | 18:18 |
| ayoung | dolphm, regarding https://review.openstack.org/#/c/12111/ are you planning on addressing gyee's comments? His -1 keeps getting carried forward. | 18:19 |
| salv-orlando | Yes, please file a bug on Quantum with all these details | 18:20 |
| *** zing has joined #openstack-dev | 18:21 | |
| *** zaitcev has joined #openstack-dev | 18:22 | |
| mnewby | salv-orlando: attempting to use the dhcp agent in xen under dom0... | 18:23 |
| mnewby | salv-orlando: Creating the tap device to connect dnsmasq to appears to succeed (and creates the port), but the linux interface does not appear to be created | 18:25 |
| salv-orlando | just received an email from your friend who does not bother showing up on IRC | 18:25 |
| mnewby | salv-orlando: armax? | 18:25 |
| *** jcooley has joined #openstack-dev | 18:25 | |
| salv-orlando | mnewby: yes | 18:26 |
| openstackgerrit | A change was merged to openstack/nova: Enable list with no dict objects to be sorted in api samples https://review.openstack.org/13481 | 18:26 |
| salv-orlando | so you do ip link show <device> and you get an error? | 18:26 |
| *** sniperd has joined #openstack-dev | 18:26 | |
| salv-orlando | mnewby: I need to leave for dinner now - Will be back online in about 60-90 minutes | 18:27 |
| salv-orlando | ttyl | 18:27 |
| mnewby | salv-orlando: Correct. | 18:27 |
| mnewby | salv-orlando: ok, hope you have a nice meal! | 18:27 |
| zykes- | awwww | 18:27 |
| zykes- | that's evil! :p | 18:27 |
| salv-orlando | mnewby: Great everybody is having this problem today!! (see zykes-) | 18:27 |
| mnewby | ah, him too… :( | 18:27 |
| *** salv-orlando has quit IRC | 18:28 | |
| mnewby | zykes-: Are you under xen too? | 18:28 |
| mnewby | Or what environment? | 18:28 |
| zykes- | mnewby: nope, Ubuntu 12.04 | 18:29 |
| zykes- | ovs 1.4.0 | 18:29 |
| mnewby | Hmmm, that's plain wierd. | 18:29 |
| mnewby | dhcp agent works fine for me in that environment. | 18:29 |
| dolphm | ayoung: yes, definitely -- i haven't made it a priority because that review is dependent on a whole stack of underlying reviews | 18:30 |
| zykes- | dhcp agent works perfectly, l3 agent not so much ;p | 18:30 |
| mnewby | zykes-: That's really weird, because the functionality at hand is interface creation in oth cases. | 18:31 |
| *** kmartin has quit IRC | 18:32 | |
| mnewby | I'd expect both to work or neither. | 18:32 |
| *** dprince has quit IRC | 18:33 | |
| *** bencherian has quit IRC | 18:33 | |
| zykes- | https://bugs.launchpad.net/quantum/+bug/1057677 | 18:34 |
| uvirtbot | Launchpad bug 1057677 in quantum "L3 OVS interface not getting create properly." [Undecided,New] | 18:34 |
| zykes- | mnewby: does it run with some options that it doesn't otherwise + | 18:34 |
| *** hemna has quit IRC | 18:35 | |
| *** hemna has joined #openstack-dev | 18:35 | |
| *** crandquist has quit IRC | 18:35 | |
| ayoung | dolphm, ah, sorry. where should I start on the reviews? I though this was the bottom of the stack | 18:35 |
| ayoung | https://review.openstack.org/#/c/12106/6 ? | 18:36 |
| dolphm | ayoung: see the Dependencies section on that review? | 18:36 |
| dolphm | ayoung: err, yeah policies :) | 18:37 |
| mnewby | zykes-: If the command were failing it might make sense. As it succeeds, I really don't understand. Time to start digging into the ovs docs. | 18:37 |
| ayoung | yeah, I misread it. OK. getting read to go to town on Policies | 18:37 |
| zykes- | mnewby: tell me if I can help in testing | 18:37 |
| * dolphm hides | 18:37 | |
| zykes- | or aiding | 18:37 |
| mnewby | zykes-: will do. I'm going to dig around the logs and see if I can find a clue. | 18:37 |
| ayoung | dolphm, BTW, I have to give a talk to my team about V3 next week. Do you have any slides, or do I need to syn\thesize from the google doc? | 18:37 |
| mnewby | zykes-: Will keep you posted. | 18:37 |
| dolphm | ayoung: google doc + pending reviews? | 18:38 |
| dolphm | ayoung: best get to approving them :) | 18:38 |
| mordred | ayoung: patch updated | 18:38 |
| ayoung | dolphm, OK, I will plan on making up some slides from the google doc. we can incorporate into the summit presentation | 18:38 |
| dolphm | ayoung: and by google doc i mean the new markdown doc lol | 18:38 |
| *** KurtMartin has joined #openstack-dev | 18:38 | |
| ayoung | dolphm, somehow I knew you would say that | 18:38 |
| *** colinmcnamara has joined #openstack-dev | 18:38 | |
| dolphm | ayoung: depending on the state of the markdown, i plan on just projecting that in the v3 session | 18:39 |
| ayoung | mordred, thanks. I'll look shortly. I assume this change is a precursor to getting rid of the Keystone test runner? | 18:39 |
| *** bencherian has joined #openstack-dev | 18:39 | |
| ayoung | dolphm, those are nice to have handy, but I think some high level, readable at a distance slides might be a better intro | 18:39 |
| *** eglynn has joined #openstack-dev | 18:40 | |
| ayoung | I'll make it happen | 18:40 |
| dolphm | ayoung: that's how i envision the markdown spec being :P | 18:40 |
| dolphm | ayoung: at least the first few sections | 18:40 |
| mordred | ayoung: nope. I made the change getting rid of it as well | 18:40 |
| ayoung | dolphm, BTW, what do you think of the idea of multiple setup_*py files, one for each sub project of keystone | 18:40 |
| dolphm | ayoung: what's new + api conventions | 18:40 |
| ayoung | with setup.py just fronting all of them | 18:40 |
| ayoung | so something like | 18:40 |
| dolphm | ayoung: sub project? | 18:40 |
| ayoung | setup_server.py | 18:40 |
| ayoung | setup_middleware.py | 18:40 |
| ayoung | setup_common.py | 18:40 |
| dolphm | ayoung: oh instead of setup.py? | 18:41 |
| ayoung | dolphm, in addition to. | 18:41 |
| ayoung | setup.py would call them | 18:41 |
| ayoung | setup.py would assume you are installing everything on one machine | 18:41 |
| ayoung | but if you want just auth_token you run | 18:41 |
| ayoung | python setup_middleware.py install | 18:41 |
| dolphm | ayoung: are we breaking apart the python packages? | 18:41 |
| * ayoung whistles nonchalantly | 18:41 | |
| dolphm | ayoung: i.e. what would python setup_middlware.py install actually install? | 18:42 |
| ayoung | dolphm, auth_token.py | 18:42 |
| ayoung | dolphm, but coupled with common | 18:42 |
| dolphm | ayoung: setup.py can't install modules | 18:42 |
| dolphm | afaik | 18:42 |
| ayoung | I realize that | 18:42 |
| ayoung | it can't install dependencies | 18:42 |
| dolphm | setup.py? | 18:42 |
| ayoung | dolphm, setup.py can install python code into site packages. It just checks for dependencies before it does so. | 18:43 |
| dolphm | setuptools.setup(install_requires=[ list of pypi depdencies ]) | 18:43 |
| ayoung | dolphm, right | 18:43 |
| ayoung | so for middleware, it would require keystone_common | 18:43 |
| dolphm | new package? | 18:44 |
| ayoung | yes, but built out of the same source tree | 18:44 |
| dolphm | /confused | 18:44 |
| ayoung | so if you do it from the top level setup.py you get all of them | 18:44 |
| dolphm | new top level package in keystone repo? | 18:44 |
| ayoung | or you can do them by hand | 18:44 |
| *** novas0x2a|laptop has joined #openstack-dev | 18:44 | |
| ayoung | dolphm, only thing that changes is the setup.py file | 18:44 |
| ayoung | and we add | 18:44 |
| ayoung | setup_common.py etc | 18:45 |
| ayoung | no split of the git repo | 18:45 |
| dolphm | ayoung: what do auth_tokens import statement look like, then? | 18:45 |
| ayoung | I think it will work, just trying to see if I am taking crazy pills | 18:45 |
| dolphm | i'm fine with one git repo, if that's all you're asking | 18:45 |
| ayoung | dolphm, I think it would depend on keyston_common only | 18:45 |
| dolphm | just trying to understand the python package structure | 18:45 |
| dolphm | would keystone_common exist if you did setup.py develop? | 18:46 |
| *** KurtMartin has quit IRC | 18:46 | |
| ayoung | we might be able to go more granular, with splitting cms out from passlib as suggested by someone smart during out meeting. | 18:46 |
| ayoung | dolphm, yes it would be kicked off by setup.puy | 18:46 |
| dolphm | so "import keystone_common" in auth_token would import from (in reality) the same keystone package on disk? | 18:47 |
| *** rohitk has joined #openstack-dev | 18:47 | |
| ayoung | we hack setup.py to bascially be a front end to a list of setup_<project>.py files | 18:47 |
| *** ewindisch has joined #openstack-dev | 18:47 | |
| dolphm | (this is confusing and i've never seen it done but +1 if it works) | 18:47 |
| ayoung | dolphm, I have absoluetly no idea, but since Python went and invented its own packaging scheme, I am just trying to find a way to work within its rules | 18:48 |
| ayoung | I think it would work, and have found vague references to is out there .... | 18:48 |
| *** KurtMartin has joined #openstack-dev | 18:49 | |
| rohitk | when does the stable branch get? | 18:49 |
| *** soosiechoi has joined #openstack-dev | 18:49 | |
| rohitk | get cut* | 18:49 |
| ayoung | dolphm, I'm going through the policy review. the only one that seems non-trivial is the sql backend. care to walk me through it? | 18:50 |
| mordred | ayoung, dolphm: lemme walk through that scrollback... I think I may have just paniced | 18:50 |
| mordred | yup. | 18:51 |
| mordred | so, we spent a bunch of time removing that pattern from quantum | 18:51 |
| ayoung | mordred, why | 18:51 |
| mordred | because it makes pypi interaction and tarball generation a nightmare | 18:51 |
| dolphm | ayoung: keystone/policy/backends/sql? | 18:52 |
| *** bencherian has quit IRC | 18:52 | |
| mordred | everything becomes wonky, and very custom-written to our repos | 18:52 |
| ayoung | mordred, ah, yeah, the tarball gen would suck. | 18:52 |
| ayoung | mordred, the thing is, Python is being stupid here | 18:52 |
| mordred | ayoung: and we have a bunch of things that expect standard python setup.py sdist to work as expected | 18:52 |
| ayoung | mordred, except that setup doesn't play by the rules that the rest of the universe lives by | 18:52 |
| mordred | ayoung: well... it does by the rules of python | 18:53 |
| *** Tross has quit IRC | 18:53 | |
| mordred | I agree with you, btw - it's a pile of monkies | 18:53 |
| mordred | but I'd posit that autotools based software also has a one-tarball-per-repo model as well | 18:53 |
| ayoung | mordred, except that I usually like monkeys | 18:53 |
| mordred | make distcheck is going to produce one and only one tarball | 18:53 |
| mnewby | zykes-: have you looked at /var/log/messages? | 18:53 |
| ayoung | mordred, yes, autotools doesn't have a package system built in to it | 18:54 |
| mordred | ayoung: what is it about multiple repos we are trying to avoid here? | 18:54 |
| zykes- | mnewby: nope | 18:54 |
| zykes- | is there something there? | 18:54 |
| mordred | ayoung: sure it does ... ever used slackware? :) | 18:54 |
| * zykes- cracks the whip at mordred | 18:54 | |
| ayoung | mordred, people want to be able to deploy auth_token and the associated keystone_common files it needs without shipping all of keystone. WHich they should be able to do | 18:54 |
| mnewby | zykes-: For me, the ovs-vsctl command returned 0, which would seem to indicate success. But messages shows that the command actually had errors. | 18:54 |
| zykes- | Slack ? Might as well use Ubuntu | 18:54 |
| mordred | ayoung: there is a re-write of all of this going on, called distutils2 - I'm planning on putting a dev on helping finish it | 18:54 |
| mordred | because I belive it gets more flexible/better to deal with | 18:54 |
| ayoung | I am trying to avoid having to change my git repo set up everytime someone needs to chip off a subset of the code | 18:54 |
| mordred | ayoung: well, changing a git repo is one time event that's pretty low cost, subverting expectations of how a python repo works is something that will effect everyone who pulls the code every time | 18:55 |
| *** bencherian has joined #openstack-dev | 18:55 | |
| mordred | and will increase their localized information ramp up cost | 18:55 |
| mordred | ayoung: I agree, btw, about auth_token deployment such as you have described | 18:56 |
| ayoung | mordred, that is 100% wrong backwards, and grounds for divorce | 18:56 |
| mordred | ayoung: uhoh | 18:56 |
| ayoung | changing a git repo is painful | 18:56 |
| mordred | ayoung: and here I thought we were mfeo | 18:56 |
| mordred | ayoung: explain that please? | 18:56 |
| ayoung | it breaks the domain model that should be common across all the code that uses it | 18:56 |
| ayoung | now I have to try and keep versions in sync across multiple git repos? | 18:57 |
| ayoung | shudder | 18:57 |
| mordred | why? | 18:57 |
| mordred | why do you have to do that? | 18:57 |
| mnewby | ayoung: isn't that what all of openstack already does? | 18:57 |
| mordred | this is what API's are for | 18:57 |
| mordred | keystone can already consume python-keystoneclient just by putting in a reference in tools/pip-requires | 18:57 |
| mordred | it's pretty easy | 18:57 |
| mordred | it just means we actually have to make and then use apis | 18:57 |
| *** markmcclain has quit IRC | 18:58 | |
| ayoung | mordred, except here is the realizty. You do something that requires code to act the same on both sides of a pipeline. Hashing, crypto, etc, so you make common code to do that. The common code is a helper, and should not be a public API. now every API is public. | 18:58 |
| ayoung | mordred, never use the word "just" when talking about software. Nothing is "just" easy... | 18:58 |
| mordred | if it needs to be done on both sides of the pipeline, then it's a de facto api | 18:58 |
| mordred | ayoung: agree re just | 18:58 |
| ayoung | mordred, no when it is talking to utility functions it isn't | 18:59 |
| mordred | yes, it is | 18:59 |
| mordred | because the data format is a defacto part of the message | 18:59 |
| *** bencherian has quit IRC | 18:59 | |
| mordred | so either there is a way to unpack that data that's consistent on both sides of the pipeline | 18:59 |
| mordred | which would be a form of contract | 18:59 |
| mordred | or ... hrm, I don't know why I said either | 19:00 |
| ayoung | mordred, look at the CMS code before you stake that claim. I am using fork/openssl but that should and can be done using an embedded library. I want it to behave the same in both places to keep code maintainence down, but making it a public API is a horror from Arkham | 19:00 |
| *** mattstep has quit IRC | 19:00 | |
| mordred | ayoung: please expand CMS? | 19:00 |
| ayoung | but Crypto Message syntax | 19:00 |
| mordred | ayoung: thank you | 19:00 |
| ayoung | the format used for PKI tokens | 19:01 |
| *** mattstep has joined #openstack-dev | 19:01 | |
| *** mattstep has quit IRC | 19:01 | |
| mordred | ayoung: why is it a horror from Arkham? | 19:01 |
| zykes- | jgriffith: is thera a way I can have cinder not shred volumes on delete ? | 19:01 |
| zykes- | it takes lots o time | 19:01 |
| *** mattstep has joined #openstack-dev | 19:01 | |
| mordred | ayoung: (not trolling, btw, in case I'm coming across that way) | 19:01 |
| ayoung | mordred, because it means that things that are supposed to be internal, and helpers, become external, and thes carved in stone. It means your proejct cannot evolve. It makes it impossible to make helper functions. It makes my life harder. That is not the goal here. The goal is and always should be to make my life easier. | 19:02 |
| mordred | ok. | 19:02 |
| mordred | so, sake of argument... | 19:02 |
| ayoung | mordred, I hope you can tell by my tone that I am not overly offended here. | 19:02 |
| mnewby | internal helpers => plugins? | 19:03 |
| mordred | ayoung: yes - it's just always good to double-check some times | 19:03 |
| mordred | ayoung: sake of argument... | 19:03 |
| *** GheRivero has quit IRC | 19:03 | |
| ayoung | mordred, the short of it is the Keystone is still a young project, and It needs time to bake. I don't want to commit to public APIs any more than necessary | 19:03 |
| *** colinmcnamara has quit IRC | 19:03 | |
| mordred | why not put those functions into python-keystoneclient somewhere, but _ prefix them | 19:03 |
| mordred | that way they can be accessible in the library, but are marked as private impl details | 19:03 |
| dolphm | mordred: that was my suggestion yesterday, i think | 19:04 |
| ayoung | mordred, from our workflow, that means I need to make changes to to both sides and keep them in sync. That is labor intensive | 19:04 |
| dolphm | mordred: have "everything" consume keystoneclient (keystone and auth_token) | 19:04 |
| mordred | dolphm: yes. | 19:04 |
| ayoung | Keystone is the grouping, I'd like to keep things internal to keystone, well, internal | 19:04 |
| mordred | ayoung: hrm. | 19:04 |
| dolphm | ayoung: remember, keystoneclient is really just a python library | 19:05 |
| ayoung | For example, one reason I modified auth_token in the first place was to make sure that the changes could work even if people only updated their keystone code, but not, say Nova or glance | 19:05 |
| dolphm | ayoung: it's a good place to put common utility functions | 19:05 |
| ayoung | we couldn't tell people "ok, now you have to change the middleware line of your paste file." | 19:05 |
| *** dubsquared has joined #openstack-dev | 19:06 | |
| dolphm | in fact, it should be called keystonelib and the CLI should be entirely provided by openstackclient | 19:06 |
| *** soosiechoi has left #openstack-dev | 19:06 | |
| mordred | dolphm: well yeah, but that's a whole other thing | 19:06 |
| mordred | :) | 19:06 |
| dolphm | :) | 19:06 |
| ayoung | dolphm, I really don't like having the CLI in a seperate project either. We are being forced to split git repos due to an external too,l, and that is the wropng abstraction. | 19:06 |
| dolphm | ayoung: sure we can | 19:06 |
| *** brianr has quit IRC | 19:06 | |
| mordred | which external tool is forcing this? | 19:06 |
| *** brianr has joined #openstack-dev | 19:06 | |
| ayoung | setup utils | 19:06 |
| mordred | that's not an external tol | 19:06 |
| mordred | tool | 19:06 |
| mordred | that's a very standard part of python development | 19:07 |
| ayoung | mordred, nah, you are just saying that because you've always used it. It is an upstart | 19:07 |
| mordred | ayoung: I have certainly not always used it :) | 19:07 |
| ayoung | mordred, but you yourself said you want to extend it | 19:07 |
| dolphm | ayoung: we've made enough changes to keystone.conf that most people are going to have to refer to keystone.conf.sample anyway | 19:07 |
| ayoung | dolphm, but that doesn't affect the middleware line ith nova's paste files | 19:07 |
| mordred | ayoung: I want to help the project that aims to replace it with a new standard | 19:07 |
| ayoung | in | 19:07 |
| dolphm | ayoung: true | 19:08 |
| ayoung | mordred, because the existing one is too limited | 19:08 |
| ayoung | so we are in violent agreement. | 19:08 |
| mordred | ayoung: I use "standard" in quotes here, but distutils (in the core) has the same limitation | 19:08 |
| mordred | ayoung: to a point - but I do not want to ad-hoc invent our own thing before we have upstream thing usable | 19:08 |
| mordred | because it took me close to two years to remove our very simple fork of nose from nova | 19:09 |
| ayoung | mordred, so people were working around this by just copying around the auth_token.py file, which they could get away with since it was a single file. But that was avoiding the use of setup.py as well. | 19:09 |
| *** brianr has quit IRC | 19:09 | |
| ayoung | mordred, I hear you. | 19:09 |
| *** brianr has joined #openstack-dev | 19:09 | |
| mordred | ayoung: yes, this is why I support putting auth_token into its own thing :) | 19:09 |
| ayoung | That is why I was floating the idea before implementing it. | 19:09 |
| mordred | ayoung: this is one of those areas which basically just causes gout | 19:09 |
| ayoung | mordred, so right now auth_token shares code with keystone server. | 19:09 |
| zykes- | mnewby: got any findings ? | 19:10 |
| mordred | ayoung: you know what we should do ... you get rh to pony up a person, I'll get a person from HP, and we'll have them just finish distutils2 so we can get to some sanity | 19:10 |
| ayoung | and instead of copying the code to both places, it lives in keystone common right now | 19:10 |
| *** dprince has joined #openstack-dev | 19:10 | |
| mordred | ayoung: with concerted effort, I bet we could have something sane in a couple of months | 19:10 |
| ayoung | mordred, we can just call it openstack common and have markmc do it! | 19:11 |
| * ayoung notes that he is not on IRC ATM | 19:11 | |
| mnewby | zykes-: As I mentioned, the ovs-vsctl command was succeeding, but under the covers adding of the tap device as a port was failing. | 19:11 |
| *** brianr has quit IRC | 19:11 | |
| mordred | ayoung: hahaha | 19:11 |
| *** brianr has joined #openstack-dev | 19:12 | |
| mordred | ayoung: (I actually have a guy starting in 10 days who I've told I'm going to force to fix distutils2 ... so I'm not talking too much out of thin air... but help is always appreciated!) | 19:12 |
| ayoung | mordred, we have more than a token investment in python here, I am sure that we can get some attention on a decent distutils approach, but I suspect that it will be too late for poor old auth token | 19:13 |
| *** sniperd has quit IRC | 19:13 | |
| ayoung | dolphm, do you really thinkg that auth_token belongs in keystone common ? | 19:14 |
| dolphm | ayoung: no? | 19:14 |
| ayoung | er, keystoneclient | 19:14 |
| ayoung | dolphm, | 19:14 |
| dolphm | ayoung: yes, or it's own repo -- but keystoneclient seems convenient | 19:14 |
| zykes- | mnewby: what was the error ? | 19:14 |
| mordred | I'm not sure about "belong" - but I think that keystoneclient is an easier place to put it, because you can't really use auth_token without keystoneclient | 19:14 |
| ayoung | mordred, not correct | 19:15 |
| ayoung | you don't need keystone client | 19:15 |
| ayoung | auth_token stands alone | 19:15 |
| ayoung | but... | 19:15 |
| mnewby | zykes-: Invalid argument | 19:15 |
| zykes- | hmmm, mnewby rootwrap error ? | 19:15 |
| ayoung | as dolphm points out, it does tend to duplicate some of the functionality | 19:15 |
| zykes- | or ovs error ? | 19:15 |
| dolphm | ayoung: my preference is really for it to consume keystoneclient, the rest is just details to me | 19:15 |
| mnewby | zykes-: But no further indication. Have to turn up the log level on ovs-vswitchd i guess. | 19:15 |
| ayoung | dolphm, OK, lets make *that* the bug | 19:15 |
| ayoung | and then I think I can get behind it | 19:16 |
| mnewby | zykes-: It's not a quantum issue. The ovs command quantum executes is against ovs-vsctl, and it succeeds. | 19:16 |
| dolphm | mordred: ayoung: long term, auth_token *needs* to be re-written to actually use keystoneclient's python API | 19:16 |
| mordred | ++ | 19:16 |
| mnewby | ++ | 19:16 |
| ayoung | dolphm, and I repeate OK, lets make *that* the bug | 19:16 |
| dolphm | blueprint? | 19:16 |
| zykes- | mnewby: tell me this then why my manual command didn't fail ? | 19:17 |
| mnewby | zykes-: was it ovs-vsctl that you ran? | 19:17 |
| *** armaan has joined #openstack-dev | 19:17 | |
| ayoung | dolphm, does it need a blueprint? I'd say that it is really just a bug: duplicated code between auth_token and keystoneclient. | 19:18 |
| zykes- | add-port command mnewby | 19:19 |
| mnewby | zykes-: Please paste whole command. | 19:19 |
| *** zing has quit IRC | 19:19 | |
| dolphm | ayoung: i have a hard time classifying duplicated code as a bug, especially considering it's redeveloping major parts of auth_token from the ground up | 19:20 |
| mordred | wait - we'd going to make a command that ports mnewby? awesome | 19:20 |
| mnewby | i'm afraid i'm not portable :p | 19:20 |
| * mordred starts a new project to attempt to port mnewby | 19:22 | |
| * mnewby wonders what exactly i'm going to be ported to | 19:23 | |
| * mnewby hopes it's silicon! | 19:23 | |
| zykes- | mnewby: https://bugs.launchpad.net/quantum/+bug/1057677 | 19:24 |
| uvirtbot | Launchpad bug 1057677 in quantum "L3 OVS interface not getting create properly." [Undecided,New] | 19:24 |
| zykes- | You see the command in there | 19:24 |
| mnewby | zykes-: Yup. If your problem is anything like mine I'm guessing ovs-vswitchd is choking on something. | 19:25 |
| mnewby | zykes-: You're installing using devstack? or…? | 19:25 |
| *** s0mik has quit IRC | 19:25 | |
| *** sniperd has joined #openstack-dev | 19:26 | |
| zykes- | mnewby: uhm, you know that command didn't give me an error right ? | 19:26 |
| mnewby | zykes-: You know my command didn't either? | 19:27 |
| mnewby | zykes-: From what I can tell ovs-vsctl makes an async call. | 19:27 |
| mnewby | zykes-: If returns success if it dispatched correctly. | 19:27 |
| zykes- | ok | 19:27 |
| mnewby | zykes-: But if there was a problem post-dispatch, you won't see it unless logging is correctly configured for ovs-vswitchd. | 19:27 |
| mnewby | zykes-: I'm going to try to figure out how to increase the logging level. Either it's a config issue or the version of ovs (it's 1.0.99 on xcp 1.5). | 19:28 |
| mnewby | zykes-: devstack? | 19:29 |
| mnewby | zykes-: The reason I'm asking is that I've used devstack on 12.04 with dhcp and l3 no issue. | 19:29 |
| mnewby | zykes-: If you're configuring other than devstack, you may want to ensure you are configuring it similarly to how it's done for devstack, since that's known to work. | 19:30 |
| *** sniperd has quit IRC | 19:30 | |
| zykes- | not devstack no mnewby | 19:32 |
| openstackgerrit | A change was merged to openstack/glance: Add test for image_tag_set_all https://review.openstack.org/13436 | 19:32 |
| mnewby | zykes-: Then there are two avenues I would suggest. | 19:32 |
| mnewby | zykes-: First - figure out how to increase the logging from ovs-vswitchd. | 19:32 |
| mnewby | zykes-: Actually, that's my only recommendation. If you can get a useful error message you'll know what to do. | 19:32 |
| *** s0mik has joined #openstack-dev | 19:34 | |
| *** jcooley has quit IRC | 19:34 | |
| dolphm | ayoung: you asked about policy sql earlier -- did you have a question on it? | 19:35 |
| *** danwent has quit IRC | 19:40 | |
| *** sniperd has joined #openstack-dev | 19:40 | |
| *** bencherian has joined #openstack-dev | 19:42 | |
| *** rohitk has quit IRC | 19:42 | |
| *** mattstep has quit IRC | 19:46 | |
| *** rohitk has joined #openstack-dev | 19:46 | |
| *** mattstep has joined #openstack-dev | 19:46 | |
| ayoung | dolphm, just returning to that. Yes | 19:46 |
| dolphm | ayoung: hit me | 19:47 |
| ayoung | dolphm, actually, now that I look at it, it seems quite simple. Seems like a lot of lines of code for fairly generic work | 19:48 |
| *** mattstep has quit IRC | 19:48 | |
| ayoung | only update is tricky, huh | 19:48 |
| dolphm | ayoung: wouldn't it be nice if we abstracted all this basic sql crud ... | 19:48 |
| *** mattstep has joined #openstack-dev | 19:48 | |
| *** mattstep has quit IRC | 19:48 | |
| ayoung | what is @handle_conflicts(type='policy') | 19:48 |
| dolphm | ayoung: defined at the top of the file | 19:48 |
| *** mattstep has joined #openstack-dev | 19:49 | |
| ayoung | dolphm ah, ok | 19:49 |
| ayoung | dolphm, OK, I feel like I should do a deeper code review of this, but it seems pretty simple to me | 19:49 |
| ayoung | is there anything in the policy patch you think is tricky or should be deeper inspected? | 19:50 |
| dolphm | ayoung: honestly, the most debatable aspect in my opinion is what the sql table looks like | 19:50 |
| dolphm | ayoung: mostly cause i don't want to write migrations to change it later :P | 19:50 |
| ayoung | dolphm, OK...lets look | 19:50 |
| ayoung | id = sql.Column(sql.String(64), primary_key=True) | 19:51 |
| *** n0ano has quit IRC | 19:51 | |
| dolphm | ayoung: ah, and there's the API-level issue of what policies attach to -- services or endpoints? | 19:51 |
| ayoung | dolphm, I take it a standard hash | 19:51 |
| ayoung | dolphm, I would say to services, but then allow override for endpoints | 19:51 |
| ayoung | by default, it should be per service | 19:51 |
| ayoung | dolphm, otherwise, they won't scale | 19:52 |
| dolphm | ayoung: at the rest api? | 19:52 |
| ayoung | dolphm, scale in the maintainance sense. I assume there needs to be a per endpoint policy mechanism, but it should be groupable | 19:53 |
| dolphm | ayoung: worse case, you're talking about 3x more admin/management calls to work with policies per service vs per endpoint based on the current impl | 19:53 |
| *** pixelbeat has quit IRC | 19:53 | |
| ayoung | dolphm, it also means that chagning a policy for all services requires a call for each, leaving the possibility of missing one... | 19:53 |
| ayoung | dolphm, do any of the openstack components have concept of "cell" or grouping? | 19:54 |
| *** s0mik has quit IRC | 19:54 | |
| dolphm | ayoung: example? | 19:55 |
| *** mmalesky has joined #openstack-dev | 19:55 | |
| *** apevec has joined #openstack-dev | 19:55 | |
| *** apevec has joined #openstack-dev | 19:55 | |
| ayoung | dolphm, well, we have tenants, so I guess that is an example. I was talking with a quantum dev, and he was trying to figure out a way to specify what set of grouping network-switches by specific hardware. | 19:56 |
| ayoung | So really I mean a way of grouping endpoints | 19:56 |
| *** sandywalsh has quit IRC | 19:56 | |
| ayoung | dolphm, so it seems to me that it should be something like policy can apply to service, endpoint, or a grouping of them. | 19:57 |
| ayoung | "all quantum servers that server macrofabric switches" for example | 19:58 |
| ayoung | or "the nova servers for the physical machines owned by the accounting department" | 19:58 |
| *** armaan has left #openstack-dev | 19:58 | |
| dolphm | ayoung: well, each endpoint has a service_id, so if you wanted to update all the policies on an endpoint, it'd be (pardon the nested for loops) "policy.update(blob=new_policy) for policy in policies.list(endpoint=endpoint) for endpoint in endpoints.list(service=service)" | 19:58 |
| ayoung | dolphm, OK, I think that we can say "record it at the endpoint" but lets provide a mechanism for deploying it according to some rules. | 19:59 |
| ayoung | so it is probably OK as is | 19:59 |
| dolphm | ayoung: what would you suggest? | 20:00 |
| ayoung | dolphm, why both blob and extra | 20:00 |
| ayoung | sh | 20:00 |
| ayoung | ah | 20:00 |
| dolphm | ayoung: blob is a string -- which in openstack's case, will usually be a json blob with a policy definition | 20:00 |
| dolphm | ayoung: extra is any other non-indexed attribute | 20:00 |
| ayoung | the extra is the ability to cram more attributes on to the object. Why do you hate the relational-model so? | 20:00 |
| ayoung | dolphm, I hate extra | 20:01 |
| ayoung | I extra hate extra | 20:01 |
| dolphm | ayoung: meh | 20:01 |
| ayoung | extra is for people that like to break things | 20:01 |
| ayoung | extra casues bugs | 20:01 |
| dolphm | ayoung: i hate when things are in extra that shouldn't be | 20:01 |
| *** zbitter has quit IRC | 20:01 | |
| ayoung | dolphm, I agree with that statment, and posit that nothing should be in extra. BUt I am feeling quarellsome today. I won't nack the review due to that | 20:02 |
| ayoung | dolphm, does it need enable/disable? That is typically in extra, and it has caused problems elsewhere? | 20:02 |
| dolphm | ayoung: uhh... enable/disable isn't part of the spec today | 20:03 |
| ayoung | dolphm, I realize. But it has come up with tenants and users | 20:03 |
| dolphm | ayoung: i think those are logically disable-able (at least users) | 20:03 |
| dolphm | ayoung: why would you want to disable a policy, and what would you expect the outcome to be? | 20:03 |
| ayoung | dolphm, um, that is rhetorical, no? | 20:04 |
| ayoung | dolphm, you diable it because it is wrong | 20:04 |
| vishy | rmk: ping | 20:04 |
| dolphm | ayoung: and then what? | 20:04 |
| ayoung | and the outcome would be that it is not applied | 20:04 |
| ayoung | usually the default is "deny everything" | 20:04 |
| dolphm | ayoung: you'd have to restart any services that are consuming it? | 20:04 |
| dolphm | ayoung: and then they fire up without a defined policy, and yeah, deny all | 20:05 |
| *** ogelbukh has quit IRC | 20:05 | |
| openstackgerrit | A change was merged to openstack/glance: Remove unused animation module https://review.openstack.org/13052 | 20:05 |
| ayoung | dolphm, sorry for being so late to this party. I realize this discussion should have happened months ago. | 20:05 |
| dolphm | ayoung: it seems like the correct approach would be to fix the policy (PATCH) and then restart any consuming services | 20:05 |
| dolphm | ayoung: no worries | 20:05 |
| ayoung | dolphm, so, disable would be to remove the line from the database? | 20:05 |
| *** salv-orlando has joined #openstack-dev | 20:06 | |
| rmk | vishy: pong | 20:06 |
| dolphm | ayoung: DELETE /policies/{policy_id} hits delete_policy() which issues a sql DELETE | 20:06 |
| *** maurosr has quit IRC | 20:06 | |
| ayoung | dolphm that makes it kindof hard to audit, thought. | 20:07 |
| dolphm | ayoung: well disable isn't really intended to be a soft-delete either | 20:07 |
| *** maurosr has joined #openstack-dev | 20:07 | |
| dolphm | ayoung: blueprint adding a deleted_at = sql.Column(sql.DateTime()) to all sql models, and updating all sql drivers to filter by "deleted_at is null" | 20:08 |
| dolphm | ? | 20:08 |
| vishy | rmk: so detach volume from stopped domain goes boom | 20:08 |
| dolphm | ayoung: and changing delete methods to soft delete | 20:08 |
| rmk | vishy: Not surprising. Easy to fix though. | 20:09 |
| rmk | vishy: Want me to take it? | 20:09 |
| rmk | Without even looking I'm going to guess the detach operation is trying to work with a live domain rather than passing the appropriate flag to modify only the running XML | 20:10 |
| vishy | rmk: https://bugs.launchpad.net/nova/+bug/1057730 | 20:10 |
| uvirtbot | Launchpad bug 1057730 in nova "libvirt: cannot detach volume from stopped domain" [High,In progress] | 20:10 |
| vishy | i'm working on it | 20:10 |
| rmk | ok | 20:10 |
| vishy | there is some weird logic that redefines the domain with a note about xml | 20:10 |
| vishy | that is quite odd | 20:10 |
| vishy | from danms | 20:10 |
| vishy | once libvirt has support for LXC hotplug, | 20:11 |
| vishy | i don't know why that makes any difference | 20:11 |
| *** n0ano has joined #openstack-dev | 20:11 | |
| *** eharney has joined #openstack-dev | 20:14 | |
| ayoung | dolphm, I'm almost tempted to say lets approve it as is, and move on, but leave that as a question to answer before we go live. But that will mean writing SQL migrate scripts | 20:16 |
| zykes- | salv-orlando: back ? | 20:16 |
| dolphm | ayoung: soft deletes? | 20:16 |
| dolphm | ayoung: or adding a disabled attribute? | 20:16 |
| *** dprince has quit IRC | 20:16 | |
| *** ogelbukh has joined #openstack-dev | 20:16 | |
| ayoung | dolphm, yeah, I like the soft-deletes approach, and it can then implement disabled for us | 20:16 |
| vishy | rmk: got it | 20:16 |
| dolphm | ayoung: i wouldn't want to expose undeleting something on the API though | 20:17 |
| ayoung | but we can, I think, approve as is, so long as we have that discussion later. | 20:17 |
| dolphm | ayoung: if something is soft deleted, i wouldn't want keystone to be able to get it back out of the db | 20:17 |
| dolphm | period | 20:17 |
| ayoung | dolphm, so long as you could somehow read it for audit, or to recreate it | 20:17 |
| dolphm | ayoung: yeah, it's up to a sql admin at that point | 20:17 |
| dolphm | set deleted_at = null | 20:18 |
| ayoung | dolphm, or a "deleted-policy" API | 20:18 |
| dolphm | no! it's evicted from keystone | 20:18 |
| mnewby | salv-orlando: *ping* | 20:18 |
| zykes- | he's hiding mnewby ;p | 20:19 |
| salv-orlando | zykes-, mnewby: here | 20:20 |
| ayoung | dolphm, OK, lets agree to have that discussion either at or after the summit. I think the design as is is OK to move forward. disable/soft delete shouldn;t hold up the rest of the work | 20:20 |
| zykes- | ooh | 20:20 |
| zykes- | speak of the devil :o | 20:21 |
| mnewby | salv-orlando: I think my problem is the version of ovs... | 20:21 |
| *** sergiutoderascu has quit IRC | 20:21 | |
| mnewby | salv-orlando: ancient - 1.0.99 | 20:21 |
| salv-orlando | mnewby: if you're running XS 6.0.2 dom 0 yes | 20:22 |
| zykes- | salv-orlando: I got 1.4.0 | 20:22 |
| mnewby | salv-orlando: I see in the logs that ovs-vswitchd is reporting an 'invalid argument'... | 20:22 |
| zykes- | so how can it be ? | 20:22 |
| zykes- | ;p | 20:22 |
| salv-orlando | mnewby: perhaps an upgrade can solve your issue | 20:22 |
| salv-orlando | zykes-: my question is… | 20:22 |
| *** bhuvan has joined #openstack-dev | 20:22 | |
| mnewby | salv-orlando: recommended docs for the upgrade? Is INSTALL.XenServer in the ovs tree good for XCP 1.5 to the best of your knowledge? | 20:23 |
| salv-orlando | the dhcp agent works fine doesn't it? (this is for zykes-) | 20:23 |
| zykes- | salv-orlando: that works dandy | 20:23 |
| salv-orlando | mnewby: it should be. But the conditional is a must in this case | 20:24 |
| salv-orlando | zykes-: ok. Gimme 5 minutes to run some checks. | 20:24 |
| dolphm | ayoung: cool | 20:24 |
| zykes- | ok :) | 20:24 |
| *** Tross has joined #openstack-dev | 20:26 | |
| salv-orlando | zykes-: if the dhcpagent works fine, then we can exclude a problem with internal interfaces. You're now running without namespaces. Can you still do a ip netns list? The list should be empty, but still better to make sure. | 20:26 |
| zykes- | empty | 20:28 |
| salv-orlando | great. Can you post the output of sudo ovs-vsctl list interface <your-external-gateway> | 20:29 |
| zykes- | http://paste.ubuntu.com/1231133/ | 20:30 |
| *** rohitk has quit IRC | 20:32 | |
| salv-orlando | Sorry I meant the qg-xxxxx interface | 20:33 |
| zykes- | http://paste.ubuntu.com/1231142/ | 20:33 |
| *** dhellmann has quit IRC | 20:33 | |
| zykes- | http://paste.ubuntu.com/1231143/ < the one i created manually | 20:33 |
| *** maurosr is now known as murilovf | 20:37 | |
| salv-orlando | I am tempted to say that your br-ex is possessed by an evil spirit | 20:37 |
| *** murilovf is now known as maurosr | 20:37 | |
| *** koolhead17 has quit IRC | 20:38 | |
| *** metral has quit IRC | 20:39 | |
| zykes- | haha :p | 20:40 |
| salv-orlando | those last two posts are all that I need - I hope. I will try to reproduce and find the root cause. | 20:41 |
| zykes- | ok, thanks very much | 20:41 |
| zykes- | can you update the bug with updates ? | 20:41 |
| salv-orlando | Out of curiosity - the l3 agent is the only quantum process running on that node, isn't it | 20:41 |
| salv-orlando | Yes I will post updates on the bug you've reported | 20:41 |
| zykes- | l3 and dhcp agent | 20:41 |
| zykes- | runs on the same node + the ovs agent | 20:41 |
| *** dubsquared has quit IRC | 20:44 | |
| salv-orlando | ok - thanks | 20:44 |
| *** s0mik has joined #openstack-dev | 20:50 | |
| *** lloydde has quit IRC | 20:51 | |
| *** adjohn has quit IRC | 20:53 | |
| *** adjohn has joined #openstack-dev | 20:53 | |
| *** andrewbogott is now known as andrewbogott_afk | 20:54 | |
| *** s0mik has quit IRC | 20:54 | |
| *** bencherian has quit IRC | 20:56 | |
| *** sniperd has quit IRC | 20:56 | |
| *** sniperd has joined #openstack-dev | 20:56 | |
| *** milner has joined #openstack-dev | 20:58 | |
| salv-orlando | zykes-: can you post your l3 agent ini file as well? | 20:59 |
| *** edygarcia has joined #openstack-dev | 20:59 | |
| *** markvoelker has quit IRC | 20:59 | |
| *** mlavalle has left #openstack-dev | 21:00 | |
| *** cp16net is now known as cp16net|away | 21:02 | |
| *** mlavalle has joined #openstack-dev | 21:03 | |
| *** danwent has joined #openstack-dev | 21:05 | |
| *** tgall_foo has quit IRC | 21:07 | |
| *** danwent has quit IRC | 21:09 | |
| *** maurosr has quit IRC | 21:09 | |
| openstackgerrit | A change was merged to openstack/openstack-common: Add exchange_name parameter to TopicConsumer https://review.openstack.org/13715 | 21:09 |
| vishy | rmk: https://review.openstack.org/13788 https://review.openstack.org/13790 https://review.openstack.org/13791 | 21:12 |
| openstackgerrit | A change was merged to openstack/openstack-common: Add the rpc service and delete manager https://review.openstack.org/11065 | 21:13 |
| openstackgerrit | A change was merged to openstack/openstack-common: Integrate eventlet backdoor https://review.openstack.org/11066 | 21:13 |
| *** andrewbogott_afk is now known as andrewbogott | 21:14 | |
| openstackgerrit | A change was merged to openstack/openstack-common: Log config on startup https://review.openstack.org/11067 | 21:14 |
| *** rkukura has quit IRC | 21:16 | |
| *** jemartin has quit IRC | 21:17 | |
| *** jemartin has joined #openstack-dev | 21:18 | |
| *** martine has quit IRC | 21:18 | |
| *** cp16net|away is now known as cp16net | 21:20 | |
| *** AlanClark has quit IRC | 21:20 | |
| *** aclark_ has joined #openstack-dev | 21:21 | |
| *** bencherian has joined #openstack-dev | 21:22 | |
| openstackgerrit | A change was merged to openstack/nova: Pass block_device_info to destroy in revert_resize https://review.openstack.org/13773 | 21:23 |
| *** hemna has quit IRC | 21:23 | |
| *** agonella has joined #openstack-dev | 21:24 | |
| *** Ruetobas has quit IRC | 21:25 | |
| *** agonella has quit IRC | 21:25 | |
| *** Ruetobas has joined #openstack-dev | 21:32 | |
| *** dachary has quit IRC | 21:35 | |
| eglynn | markwash: there? | 21:36 |
| eglynn | markwash: just noticed a small wrinkle in http://summit.openstack.org/cfp/details/17 | 21:36 |
| *** galthaus has joined #openstack-dev | 21:36 | |
| eglynn | markwash: copy-from is already async in Folsom ... https://review.openstack.org/12438 | 21:37 |
| *** dachary has joined #openstack-dev | 21:38 | |
| *** tgall_foo has joined #openstack-dev | 21:40 | |
| *** tgall_foo has quit IRC | 21:40 | |
| *** tgall_foo has joined #openstack-dev | 21:40 | |
| markwash | eglynn: actually I was thinking of something a little broader | 21:41 |
| markwash | eglynn: that would move that kind of asyncronous operation out of the api node | 21:41 |
| eglynn | markwash: a-ha, ok, gotcha | 21:41 |
| markwash | eglynn: its basically a prelude to more intensive stuff, like image conversion operations | 21:42 |
| markwash | eglynn: which is not to say that it necessarily warrants a slot at the conference! | 21:42 |
| *** aclark_ has quit IRC | 21:42 | |
| *** littleidea has quit IRC | 21:43 | |
| eglynn | markwash: cool ... BTW I liked your incremental imaging proposal, interesting! | 21:43 |
| *** dubsquared has joined #openstack-dev | 21:43 | |
| markwash | eglynn: thanks, I'm glad. . I'm hoping we can come up with something people will like there | 21:44 |
| *** dspano has quit IRC | 21:48 | |
| *** alobbs has quit IRC | 21:50 | |
| *** rhagarty has joined #openstack-dev | 21:50 | |
| *** mtreinish has quit IRC | 21:51 | |
| *** kbringard has quit IRC | 21:52 | |
| *** markmcclain has joined #openstack-dev | 21:55 | |
| *** datsun180b has quit IRC | 21:55 | |
| *** e1mer has joined #openstack-dev | 21:58 | |
| *** hazmat has quit IRC | 21:58 | |
| *** milner has quit IRC | 22:00 | |
| *** ayoung has quit IRC | 22:03 | |
| *** pixelbeat has joined #openstack-dev | 22:03 | |
| *** metral has joined #openstack-dev | 22:03 | |
| *** metral has quit IRC | 22:04 | |
| openstackgerrit | A change was merged to openstack/glance: Add test for db api method image_member_create https://review.openstack.org/13440 | 22:06 |
| *** alunduil has quit IRC | 22:07 | |
| *** dolphm has quit IRC | 22:08 | |
| *** mrodden has quit IRC | 22:09 | |
| *** bencherian has quit IRC | 22:12 | |
| *** mnewby has quit IRC | 22:16 | |
| *** galthaus has left #openstack-dev | 22:22 | |
| rmk | vishy: Sorry had to step away for a while. I'll check those out now. | 22:22 |
| *** edygarcia has quit IRC | 22:24 | |
| *** gatuus has quit IRC | 22:26 | |
| *** mrodden has joined #openstack-dev | 22:26 | |
| *** Gordonz has quit IRC | 22:27 | |
| *** mnewby has joined #openstack-dev | 22:30 | |
| vishy | well i used & instead of or | :( | 22:36 |
| vishy | which works but I don't think it is exactly what i want | 22:36 |
| *** rnirmal has quit IRC | 22:39 | |
| *** pixelbeat has quit IRC | 22:42 | |
| *** roge has quit IRC | 22:43 | |
| *** sniperd has joined #openstack-dev | 22:46 | |
| *** jimfehlig has quit IRC | 22:52 | |
| *** mnewby has quit IRC | 22:53 | |
| *** danwent has joined #openstack-dev | 22:54 | |
| *** s0mik has joined #openstack-dev | 22:55 | |
| *** markmcclain has quit IRC | 22:56 | |
| *** markmcclain has joined #openstack-dev | 22:56 | |
| *** dubsquared has quit IRC | 22:57 | |
| *** eharney has quit IRC | 22:59 | |
| *** markmcclain has quit IRC | 22:59 | |
| *** jcooley has joined #openstack-dev | 22:59 | |
| *** markmcclain has joined #openstack-dev | 23:00 | |
| *** adjohn has quit IRC | 23:00 | |
| *** dubsquared has joined #openstack-dev | 23:00 | |
| *** titankiller has quit IRC | 23:00 | |
| *** dubsquared has quit IRC | 23:03 | |
| *** littleidea has joined #openstack-dev | 23:05 | |
| *** littleidea has joined #openstack-dev | 23:06 | |
| vishy | danwent: ping | 23:07 |
| danwent | vishy: pong | 23:07 |
| vishy | danwent: two things, I wasn't sure if you ever saw that review i passed during your meeting last week | 23:07 |
| vishy | danwent: it adds calls to network during migration for floating ips | 23:08 |
| danwent | vishy: i did not, sorry, too many things scrolling on the screen. can you resend? | 23:08 |
| vishy | danwent: https://review.openstack.org/#/c/13539/ | 23:09 |
| danwent | yeah, passing for now is the right thing to do with quantum. | 23:10 |
| vishy | danwent: I don't think quantum has the concept of multihost natting right now | 23:10 |
| vishy | but I figure eventually it might have to do stuff during migrate | 23:10 |
| danwent | exactly | 23:10 |
| vishy | danwent: although perhaps it shouldn't be specific to floating ips | 23:10 |
| danwent | haha, was just going to make the same comment | 23:10 |
| vishy | should it just be migrate_instance_start? | 23:10 |
| danwent | I think there's a lot of value in a generic pre-migration + post-migration notification mechanism | 23:10 |
| vishy | agreed | 23:11 |
| danwent | that is what I would focus on | 23:11 |
| vishy | I will ask him to change it. We can just implement floating ips for now | 23:11 |
| danwent | let me add that to my list of quantum-related future topics | 23:11 |
| vishy | danwent: ok next question | 23:11 |
| danwent | shoot | 23:12 |
| vishy | so the requested networks stuff | 23:12 |
| vishy | you can pass in network_uuids, ip_addresses, and ports | 23:13 |
| vishy | I notice that quantum doesn't do anything with ip addresses | 23:13 |
| *** alunduil has joined #openstack-dev | 23:13 | |
| vishy | danwent: is the idea ever to support that? or is it port only? | 23:14 |
| danwent | there was a bug discussion of this a while back | 23:14 |
| danwent | quantum already supports creating a port with an IP, and then passing that port through nova | 23:15 |
| vishy | danwent: also, are ports uuids? | 23:15 |
| danwent | my goal was that I didn't want people proxying every god-damn network port setting through nova | 23:15 |
| danwent | since fixed-ips were already there as an option, i would be ok with quantumv2/api.py implementing that particular option | 23:15 |
| danwent | I think there was a patch for it that may have rotted on the vine, as I remember reviewing it once, but not seeing a re-rev (I may have missed it) | 23:16 |
| vishy | danwent: ok so ports uuids? or no | 23:16 |
| danwent | sorry, yes | 23:16 |
| danwent | https://bugs.launchpad.net/nova/+bug/1046358 | 23:16 |
| *** andrewbogott is now known as andrewbogott_afk | 23:16 | |
| uvirtbot | Launchpad bug 1046358 in nova "Fixed ip assignment using --nic parameter doesn't work" [High,Triaged] | 23:16 |
| vishy | danwent: I'm asking because specifying a network uuid seems kind of extra | 23:17 |
| vishy | if you already pass a port | 23:17 |
| danwent | yes, its an either or thing, I believe | 23:17 |
| vishy | so i was thinking the following. Network_id is optional | 23:17 |
| danwent | if all you care about is a network, just pass the network-id | 23:17 |
| danwent | if you care are specific settings on the port, create a port on that network, and just pass the port-id | 23:17 |
| vishy | danwent: ah it actually supports it in quantum! | 23:18 |
| danwent | ? | 23:18 |
| vishy | sorry i didn't realize you could pass port without network | 23:19 |
| vishy | great that makes it easier | 23:19 |
| *** ev0ldave has left #openstack-dev | 23:19 | |
| vishy | so, I want to support just passing ip without network as well | 23:19 |
| vishy | and if there is only one matching ip then it works | 23:19 |
| vishy | if there are duplicates due to overlapping networks it could fail | 23:19 |
| vishy | does that make sense to you? | 23:20 |
| danwent | hmmm… yeah, i could imagine that model | 23:20 |
| vishy | danwent: it would be like: grab all networks available to the user, look for networks that have the requested ip | 23:20 |
| vishy | return it if there are exactly 1 | 23:21 |
| vishy | is there a way to determine if a network contains an ip in quantum? | 23:21 |
| danwent | sure.. in quantum speak it would be all "subnets" associated with the tenant, but yeah, makes sense | 23:21 |
| vishy | danwent: I was going off of: | 23:21 |
| vishy | 66 def _get_available_networks(self, context, project_id, | 23:21 |
| vishy | 67 net_ids=None): | 23:21 |
| vishy | 68 """Return a network list available for the tenant. | 23:21 |
| danwent | vishy: yup, my only point was that in quantum networks don't have cidrs, they have subnets, which in turn have cidrs | 23:22 |
| danwent | but same basic idea | 23:22 |
| danwent | there's no explicit call to "search" subnets that contain a particular IP. | 23:22 |
| vishy | danwent: ah ok, how would you determine if the ip is in use? | 23:22 |
| danwent | you'd have to list the subnets, and then to a matching test. | 23:22 |
| danwent | its easy to see if an IP is in use, as you can search for a port with that IP | 23:23 |
| vishy | danwent: so it sounds like a bunch of calls: get all the networks, get subnets, match cidrs, make sure port isn't in use. | 23:24 |
| danwent | so you first would ahve to identify if it matches a subnet, confirm that there is only one such match, then confirm that no port already uses that IP in the subnet | 23:24 |
| vishy | danwent: but that makes sense | 23:24 |
| danwent | you could actually skip the first get all networks call | 23:24 |
| danwent | just get all subnets, do match, then search on ports. | 23:24 |
| vishy | danwent: the matching in nova is a little easier, but I didn't want to add it if quantum can't do it | 23:24 |
| danwent | so only two api calls, I think. | 23:25 |
| vishy | danwent: don't you have to verify the subnet is accessible? | 23:25 |
| danwent | you will only get subnets that are associated with that tenant when you search subnets | 23:25 |
| vishy | danwent: I guess you have some logic in the subnet search which will return tenant subnets and public subents? | 23:25 |
| vishy | * subnets | 23:26 |
| *** jcooley has quit IRC | 23:26 | |
| danwent | this is a bit tricky with subnets, as I don't think we expose shared subnets to tenants directly. The idea is that tenant's dont' really "own" those subnets, and so shouldn't be able to choose IPs out of those subnets (only have them assigned). | 23:27 |
| danwent | I'd have to check with salv-orlando to confirm though | 23:27 |
| salv-orlando | sorry guys I was focused on another screen | 23:28 |
| danwent | is your use case a tenant choosing their IP on a shared subnet, or a private one? | 23:28 |
| danwent | salv-orlando: no worries, just talking through a use case with vishy and wasn't clear on whether a shared subnet would be visible to a non-admin tenant querying all subnets | 23:28 |
| vishy | danwent: well I would think both | 23:29 |
| vishy | danwent: shared subnet could be useful for private clouds | 23:29 |
| vishy | danwent: at least in the nova-network case the shared is the important use case. With quantum it may be less so. | 23:30 |
| danwent | i guess our thinking was that "shared" networks are accessible to multiple tenants, so no tenant could really make any assumption about IPs may or may not be available for them to reserve | 23:31 |
| danwent | so reserving a specific IP didn't make sense | 23:31 |
| vishy | danwent: yes that is the case, unless they have the ability to list them | 23:31 |
| vishy | danwent: i could see private clouds where it would make sense though | 23:31 |
| salv-orlando | anyway, a non-admin tenant can see shared subnets just by listing all of them | 23:32 |
| danwent | ok, so looks like I was wrong anyway :) | 23:32 |
| vishy | danwent: say you have one network with public ips | 23:32 |
| danwent | i'm still curious about the use case though | 23:32 |
| vishy | danwent: it would be nice to let users allocate those specifically | 23:32 |
| salv-orlando | danwent: you can see the cidr, but not which IPs are allocated | 23:33 |
| danwent | salv-orlando: but can a user include fixed_ips = [ {"subnet_id": XXX, "fixed_ip": YYY}] for a shared subnet? | 23:33 |
| salv-orlando | no | 23:33 |
| danwent | ok, that's what we're talking about here | 23:33 |
| salv-orlando | that is forbidden bu policy | 23:33 |
| vishy | danwent: I guess the only advantage of advance allocation is you know ahead of time what it is | 23:33 |
| vishy | but if you can allocate one to a port and reuse the port that solves the same thing | 23:34 |
| salv-orlando | but the policy is not hardcoded; if you remove that policy from policy.json, you can do whatever you want | 23:34 |
| danwent | yeah, that was our basic model for "ip reservation" | 23:34 |
| vishy | danwent: nova-network doesn't have the concept of a port so the only way you get reserved ips is using floating ips which gets you into the realm of natting | 23:34 |
| vishy | danwent: I think the port concept makes my usecase irrelevent | 23:35 |
| *** aeperezt has quit IRC | 23:35 | |
| danwent | vishy: yup. with quantum, you should be able to get a port on a public network, which represents your allocation of an IP on that network. | 23:35 |
| danwent | I think with nova right now the only way to assign or unassign that port of a VM is by booting or deleting that vm | 23:35 |
| vishy | danwent: question, are floating ips implemented with natting? | 23:35 |
| danwent | but if vif hotplugging was supported, you could imagine them on the fly | 23:36 |
| vishy | danwent: we should add it! is there a blueprint for grizzly? | 23:36 |
| danwent | on hotplug? I think there was one for folsom, but it got bogged down and I believe never finished | 23:36 |
| vishy | danwent: It seems like you could implement reserving "floating ips" through port mechanism | 23:37 |
| danwent | vishy: yes, on floating IPs, yes, they use NAT by definition to do the mapping, but with Quantum I believe you could make it so that VMs on a "public" network also had access to floating IPs. | 23:37 |
| danwent | vishy: can you elaborate? | 23:37 |
| vishy | danwent: and if you could hotplug them then you wouldn't need to nat at all. Could replace floating ips completely | 23:37 |
| danwent | ah, i see what you mean | 23:37 |
| vishy | danwent: well people use floating ips for failover. | 23:37 |
| danwent | yes, then you could do fail-over with either mechanism. | 23:38 |
| vishy | so if you could allocate a public port and move it between vms…you get the advantage of floating without having to deal with the hassle of natting | 23:38 |
| vishy | which is fragile and breaks things | 23:38 |
| *** Ryan_Lane has joined #openstack-dev | 23:39 | |
| danwent | vishy: yes, as long as you have a networking solution that can make sure a VM can use an IP regardless of its physical location | 23:39 |
| danwent | (e.g., an overlay solution) | 23:39 |
| *** epim has quit IRC | 23:57 | |
| *** MarkAtwood has joined #openstack-dev | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!