| *** coli_ has quit IRC | 00:00 | |
| *** zul has joined #openstack-dev | 00:02 | |
| *** negronjl has quit IRC | 00:08 | |
| *** jeffjapan has joined #openstack-dev | 00:09 | |
| *** mattray has quit IRC | 00:13 | |
| *** rkukura has left #openstack-dev | 00:15 | |
| openstackgerrit | Jake Dahn proposed a change to openstack/horizon: Quantum was removed from pypi, this fixes the pip-requires. https://review.openstack.org/1933 | 00:21 |
|---|---|---|
| *** code_franco has quit IRC | 00:22 | |
| *** sleepsonthefloo has quit IRC | 00:25 | |
| *** andrewbogott_ has joined #openstack-dev | 00:26 | |
| *** andrewbogott has quit IRC | 00:26 | |
| *** andrewbogott_ is now known as andrewbogott | 00:26 | |
| *** coli has joined #openstack-dev | 00:31 | |
| openstackgerrit | Chris Behrens proposed a change to openstack/nova: Fix RPC responses to allow None response correctly. https://review.openstack.org/1934 | 00:39 |
| *** reed_ has quit IRC | 00:42 | |
| *** rkukura has joined #openstack-dev | 00:43 | |
| *** negronjl has joined #openstack-dev | 00:44 | |
| openstackgerrit | A change to openstack/quantum has been rejected: Move tools to quantum.tools https://review.openstack.org/1857 | 00:50 |
| *** zns has joined #openstack-dev | 00:50 | |
| *** zaitcev has joined #openstack-dev | 00:53 | |
| *** reed has joined #openstack-dev | 00:55 | |
| openstackgerrit | Brad Hall proposed a change to openstack/quantum: bug #891246: Fix paths in agent Makefile https://review.openstack.org/1693 | 00:56 |
| uvirtbot | Launchpad bug 891246 in quantum "The Makefile for building the openvswitch plugin agent tarball no longer works" [High,In progress] https://launchpad.net/bugs/891246 | 00:56 |
| *** AlanClark has quit IRC | 00:57 | |
| openstackgerrit | anotherjesse proposed a change to openstack/nova: removed logic of throwing exception if no floating ip https://review.openstack.org/1935 | 01:01 |
| *** ben_duyujie has quit IRC | 01:04 | |
| *** lorin1 has joined #openstack-dev | 01:11 | |
| *** anotherjesse has quit IRC | 01:12 | |
| openstackgerrit | Jay Pipes proposed a change to openstack/keystone: Fixes a number of configuration/startup bugs https://review.openstack.org/1936 | 01:13 |
| *** anotherjesse has joined #openstack-dev | 01:18 | |
| *** donaldngo_hp has quit IRC | 01:24 | |
| openstackgerrit | Jake Dahn proposed a change to openstack/horizon: adding danger class to keypair delete button https://review.openstack.org/1937 | 01:25 |
| *** anotherjesse has quit IRC | 01:25 | |
| *** jeffjapan has quit IRC | 01:27 | |
| *** jeffjapan has joined #openstack-dev | 01:29 | |
| *** hugokuo has joined #openstack-dev | 01:34 | |
| *** cp16net has quit IRC | 01:35 | |
| *** mwhooker has quit IRC | 01:35 | |
| openstackgerrit | Lorin Hochstein proposed a change to openstack/glance: Added some examples for "glance add" https://review.openstack.org/1888 | 01:36 |
| *** vladimir3p has quit IRC | 01:47 | |
| *** jdurgin has quit IRC | 01:55 | |
| *** novas0x2a|laptop has quit IRC | 02:07 | |
| *** lloydde has quit IRC | 02:11 | |
| *** bengrue has quit IRC | 02:13 | |
| *** ayoung has quit IRC | 02:18 | |
| *** dragondm has quit IRC | 02:23 | |
| *** adjohn has quit IRC | 02:25 | |
| *** reed has quit IRC | 02:26 | |
| *** pixelbeat has quit IRC | 02:38 | |
| *** andrewbogott has quit IRC | 02:45 | |
| *** andrewbogott has joined #openstack-dev | 02:48 | |
| *** negronjl has quit IRC | 02:54 | |
| *** negronjl has joined #openstack-dev | 02:55 | |
| openstackgerrit | Chris Behrens proposed a change to openstack/nova: Fix RPC responses to allow None response correctly. https://review.openstack.org/1934 | 03:00 |
| *** andrewbogott has quit IRC | 03:01 | |
| *** scottjg has joined #openstack-dev | 03:11 | |
| *** vladimir3p has joined #openstack-dev | 03:16 | |
| *** andrewbogott has joined #openstack-dev | 03:18 | |
| *** andrewbogott has quit IRC | 03:19 | |
| *** jog0 has quit IRC | 03:19 | |
| *** lorin1 has quit IRC | 03:21 | |
| *** Guest28018 has joined #openstack-dev | 03:26 | |
| *** yshh has joined #openstack-dev | 03:29 | |
| *** Guest28018 has left #openstack-dev | 03:29 | |
| *** adjohn has joined #openstack-dev | 03:36 | |
| *** heut2008 has joined #openstack-dev | 03:37 | |
| heut2008 | i wonder when using multi_host how metadata forward to nova-api | 03:37 |
| heut2008 | for now ,in my multi node with muti_host , | 03:38 |
| heut2008 | matadata cant be got by instance | 03:38 |
| *** adjohn has quit IRC | 03:40 | |
| *** lorin1 has joined #openstack-dev | 03:44 | |
| *** lorin1 has joined #openstack-dev | 03:48 | |
| *** rnirmal has joined #openstack-dev | 03:49 | |
| *** lorin1 has quit IRC | 03:53 | |
| *** donaldngo_hp has joined #openstack-dev | 04:27 | |
| *** adjohn has joined #openstack-dev | 04:27 | |
| *** mjfork has quit IRC | 04:29 | |
| *** negronjl has quit IRC | 04:40 | |
| *** negronjl has joined #openstack-dev | 04:41 | |
| *** negronjl has quit IRC | 04:46 | |
| *** negronjl has joined #openstack-dev | 04:46 | |
| *** negronjl_ has joined #openstack-dev | 04:47 | |
| *** negronjl_ has quit IRC | 04:48 | |
| *** negronjl has quit IRC | 04:50 | |
| *** negronjl has joined #openstack-dev | 04:50 | |
| *** Rajaram has joined #openstack-dev | 04:51 | |
| *** negronjl has quit IRC | 04:52 | |
| *** negronjl has joined #openstack-dev | 04:53 | |
| *** negronjl_ has joined #openstack-dev | 04:54 | |
| *** negronjl_ has quit IRC | 04:55 | |
| *** negronjl_ has joined #openstack-dev | 04:56 | |
| *** yshh has quit IRC | 04:57 | |
| *** negronjl_ has quit IRC | 04:57 | |
| *** sleepsonthefloo has joined #openstack-dev | 05:13 | |
| *** vladimir3p has quit IRC | 05:24 | |
| *** reed has joined #openstack-dev | 05:25 | |
| *** koolhead17 has joined #openstack-dev | 05:44 | |
| *** ben_duyujie has joined #openstack-dev | 05:55 | |
| *** rnirmal has quit IRC | 05:55 | |
| *** adjohn has quit IRC | 06:07 | |
| *** adjohn has joined #openstack-dev | 06:07 | |
| *** zaitcev has quit IRC | 06:13 | |
| *** beekhof has quit IRC | 06:19 | |
| *** jakedahn has quit IRC | 06:24 | |
| *** jakedahn has joined #openstack-dev | 06:29 | |
| openstackgerrit | Brad Hall proposed a change to openstack-dev/devstack: Fixes to work with Quantum trunk https://review.openstack.org/1938 | 06:37 |
| *** adjohn has quit IRC | 06:48 | |
| *** jog0 has joined #openstack-dev | 06:50 | |
| *** anotherjesse has joined #openstack-dev | 06:59 | |
| *** jakedahn has quit IRC | 07:01 | |
| openstackgerrit | Brad Hall proposed a change to openstack/nova: Fix QuantumManager update_dhcp calls https://review.openstack.org/1939 | 07:02 |
| openstackgerrit | Brad Hall proposed a change to openstack/nova: Add NAT/gateway support to QuantumManager https://review.openstack.org/1940 | 07:02 |
| *** jakedahn has joined #openstack-dev | 07:04 | |
| *** ben_duyujie has quit IRC | 07:04 | |
| *** anotherjesse has quit IRC | 07:07 | |
| *** koolhead17 has quit IRC | 07:12 | |
| *** zns has quit IRC | 07:17 | |
| *** vladimir3p has joined #openstack-dev | 07:21 | |
| *** vladimir3p has quit IRC | 07:26 | |
| *** jog0 has quit IRC | 07:30 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/quantum: Added timeout flag to ovs-vsctl to avoid infinte waiting https://review.openstack.org/1715 | 07:42 |
| openstackgerrit | Ghe Rivero proposed a change to openstack/quantum: Added timeout flag to ovs-vsctl to avoid infinte waiting https://review.openstack.org/1715 | 07:50 |
| openstackgerrit | Jimmy Bergman proposed a change to openstack/nova: Fix for EC2 API part of bug 897164 https://review.openstack.org/1916 | 07:56 |
| uvirtbot | Launchpad bug 897164 in nova "neither EC2 api nor openstack API expose the hostname" [Undecided,New] https://launchpad.net/bugs/897164 | 07:56 |
| *** popux_ has joined #openstack-dev | 08:22 | |
| *** reidrac has joined #openstack-dev | 08:24 | |
| ttx | notmyname: I only see one | 08:25 |
| ttx | (at 20:00 UTC) | 08:25 |
| ttx | notmyname: reading the ICS feed directly confirms that (two expired entries named "openstack PPB meeting", one non-expired entry named "PPB meeting") | 08:32 |
| ttx | notmyname: what are you seeing exactly ? | 08:33 |
| *** scottjg has quit IRC | 09:30 | |
| *** heut2008 has quit IRC | 09:30 | |
| *** heut2008 has joined #openstack-dev | 09:30 | |
| *** Diaoer has joined #openstack-dev | 09:31 | |
| *** jakedahn has quit IRC | 09:45 | |
| *** darraghb has joined #openstack-dev | 09:47 | |
| *** pixelbeat has joined #openstack-dev | 09:48 | |
| *** jeffjapan has quit IRC | 09:59 | |
| *** apevec has joined #openstack-dev | 10:40 | |
| *** popux_ has quit IRC | 10:44 | |
| *** ben_duyujie has joined #openstack-dev | 10:47 | |
| *** ben_duyujie has quit IRC | 10:51 | |
| *** mnour has joined #openstack-dev | 11:06 | |
| *** popux_ has joined #openstack-dev | 11:25 | |
| *** ben_duyujie has joined #openstack-dev | 11:29 | |
| *** popux_ has quit IRC | 11:55 | |
| *** bsza has joined #openstack-dev | 11:57 | |
| *** rkukura has quit IRC | 12:19 | |
| *** lorin1 has joined #openstack-dev | 12:19 | |
| *** lorin1 has quit IRC | 12:22 | |
| *** popux has joined #openstack-dev | 12:22 | |
| *** mjfork has joined #openstack-dev | 12:30 | |
| *** markvoelker has joined #openstack-dev | 12:51 | |
| *** popux has quit IRC | 12:56 | |
| *** popux has joined #openstack-dev | 12:57 | |
| *** markvoelker has quit IRC | 13:03 | |
| *** donaldngo_hp has quit IRC | 13:04 | |
| *** ben_duyujie has quit IRC | 13:11 | |
| notmyname | ttx: annegentle fixed it for me yesterday. thanks | 13:25 |
| ttx | that explains it :) | 13:25 |
| notmyname | ttx: there was a 2nd ppb meeting from somebody at piston, I think | 13:26 |
| *** markvoelker has joined #openstack-dev | 13:34 | |
| *** lts has joined #openstack-dev | 13:36 | |
| *** coli has quit IRC | 13:59 | |
| *** lorin1 has joined #openstack-dev | 14:01 | |
| *** ben_duyujie has joined #openstack-dev | 14:06 | |
| *** heut2008 has quit IRC | 14:22 | |
| *** lloydde has joined #openstack-dev | 14:24 | |
| *** rkukura has joined #openstack-dev | 14:27 | |
| annegentle | ttx: well it was a response from someone at Piston that "stuck" I guess. | 14:27 |
| lloydde | you rang? | 14:27 |
| *** lorin1 has quit IRC | 14:30 | |
| *** pringles_ has quit IRC | 14:30 | |
| zul | ec2-api team meeting in less than half hour | 14:31 |
| *** kbringard has joined #openstack-dev | 14:31 | |
| ttx | zul: is it recurrent now ? If yes, please add to http://wiki.openstack.org/Meetings | 14:33 |
| ttx | zul: Then I'll add it to the gCal. | 14:33 |
| annegentle | lloydde: heh, no, but now I know to say Piston if I need you :) | 14:34 |
| annegentle | lloydde: that or "stuck" is your keyword? | 14:34 |
| lloydde | XD | 14:34 |
| zul | ttx: done | 14:35 |
| *** lorin1 has joined #openstack-dev | 14:36 | |
| *** pringles_ has joined #openstack-dev | 14:36 | |
| ttx | zul: done | 14:36 |
| zul | ttx: thanks | 14:36 |
| *** zns has joined #openstack-dev | 14:40 | |
| *** coli has joined #openstack-dev | 15:15 | |
| openstackgerrit | Mark Washenberger proposed a change to openstack/nova: Use uuids for file injection. https://review.openstack.org/1927 | 15:15 |
| *** crobinso has joined #openstack-dev | 15:19 | |
| openstackgerrit | A change to openstack/nova has been rejected: Add userfriendly message for NotImplemented https://review.openstack.org/831 | 15:20 |
| *** ayoung has joined #openstack-dev | 15:22 | |
| openstackgerrit | Alex Meade proposed a change to openstack/nova: power_on/power_off in compute manager to use uuids https://review.openstack.org/1869 | 15:47 |
| *** code_franco has joined #openstack-dev | 15:50 | |
| *** rnirmal has joined #openstack-dev | 15:50 | |
| *** bcwaldon has joined #openstack-dev | 15:57 | |
| *** dragondm has joined #openstack-dev | 15:59 | |
| *** ayoung has quit IRC | 16:00 | |
| openstackgerrit | Mark McLoughlin proposed a change to openstack/keystone: Check that endpointTemplate ID is valid in endpoint add cmd (#897749) https://review.openstack.org/1941 | 16:01 |
| *** adjohn has joined #openstack-dev | 16:01 | |
| *** lorin1 has quit IRC | 16:03 | |
| *** lorin1 has joined #openstack-dev | 16:03 | |
| *** lorin1 has quit IRC | 16:03 | |
| *** zns has quit IRC | 16:03 | |
| *** ayoung has joined #openstack-dev | 16:03 | |
| *** zns has joined #openstack-dev | 16:03 | |
| *** Diaoer has joined #openstack-dev | 16:06 | |
| *** zns has quit IRC | 16:13 | |
| *** vladimir3p has joined #openstack-dev | 16:17 | |
| *** vladimir3p has quit IRC | 16:17 | |
| *** vladimir3p has joined #openstack-dev | 16:18 | |
| *** zns has joined #openstack-dev | 16:20 | |
| *** vladimir3p has quit IRC | 16:23 | |
| *** ben_duyujie has quit IRC | 16:24 | |
| *** vladimir3p has joined #openstack-dev | 16:34 | |
| openstackgerrit | Mark Washenberger proposed a change to openstack/nova: Use uuids for compute manager agent update. https://review.openstack.org/1942 | 16:37 |
| *** vladimir3p has quit IRC | 16:38 | |
| openstackgerrit | Brian Waldon proposed a change to openstack/nova: Convert compute manager delete methods to objects https://review.openstack.org/1928 | 16:39 |
| *** bcwaldon has quit IRC | 16:40 | |
| *** reed has joined #openstack-dev | 16:41 | |
| *** Tribaal has quit IRC | 16:41 | |
| *** reidrac has quit IRC | 16:42 | |
| *** Rajaram has quit IRC | 16:47 | |
| *** negronjl_ has joined #openstack-dev | 16:50 | |
| *** adiantum has joined #openstack-dev | 16:51 | |
| *** bcwaldon has joined #openstack-dev | 16:54 | |
| openstackgerrit | A change to openstack/nova has been rejected: Convert compute manager delete methods to objects https://review.openstack.org/1928 | 16:56 |
| *** zns has quit IRC | 16:58 | |
| *** mnour has quit IRC | 17:01 | |
| openstackgerrit | Chris Behrens proposed a change to openstack/nova: Fix RPC responses to allow None response correctly. https://review.openstack.org/1934 | 17:02 |
| *** zns has joined #openstack-dev | 17:02 | |
| *** vladimir3p has joined #openstack-dev | 17:05 | |
| *** Diaoer has quit IRC | 17:05 | |
| *** zaitcev has joined #openstack-dev | 17:07 | |
| *** zns1 has joined #openstack-dev | 17:08 | |
| *** popux has quit IRC | 17:08 | |
| *** zns has quit IRC | 17:09 | |
| *** zns1 has quit IRC | 17:10 | |
| vladimir3p | vishy: ping | 17:10 |
| *** ayoung has quit IRC | 17:10 | |
| *** zns has joined #openstack-dev | 17:12 | |
| vishy | vladimir3p: sup? | 17:13 |
| vladimir3p | vishy: Hey, wanted to consult with you on rebuild/resize when you will have a moment | 17:13 |
| *** vishy has left #openstack-dev | 17:13 | |
| kbringard | lol, guess he wasn't interested | 17:14 |
| *** vishy has joined #openstack-dev | 17:14 | |
| vladimir3p | vishy: not sure if you've seen the previous msg. So, we need to consult with you on rebuild/resize | 17:15 |
| vishy | sorry | 17:15 |
| *** jakedahn has joined #openstack-dev | 17:15 | |
| vishy | yeah had a computer 'splode | 17:15 |
| *** jakedahn has quit IRC | 17:15 | |
| *** jakedahn has joined #openstack-dev | 17:15 | |
| vishy | ok | 17:15 |
| *** koolhead17 has joined #openstack-dev | 17:15 | |
| *** Tribaal has joined #openstack-dev | 17:15 | |
| vladimir3p | vishy: np, so we are tryng to find the best way to re-spawn instances on different nodes. Somthing like rebuild, but with different host | 17:16 |
| vladimir3p | vishy: there are several use-cases: host failure, possible upgrade (new image), etc. | 17:16 |
| vishy | vladimir3p: rebuild generally chooses a different host | 17:16 |
| vishy | * resize | 17:16 |
| vladimir3p | vishy: yeah, but it is not really resize | 17:17 |
| vladimir3p | vishy: we don't really care about the previous image, only instance identity (ID, IPs, etc) | 17:17 |
| vishy | interesting | 17:17 |
| vladimir3p | vishy: also, the prevous instance might be up or down ... | 17:17 |
| vladimir3p | vishy: for example host crash - now we want to recreate all instances on other nodes | 17:18 |
| vladimir3p | vishy: it is mostly useful for "special" (admin) instances and less for user instances | 17:19 |
| vishy | vladimir3p: why can't you client side relaunch them? | 17:19 |
| vishy | vladimir3p: meaning why does it need the same id? | 17:19 |
| vladimir3p | vishy: great question :-) | 17:20 |
| vladimir3p | vishy: internally there is a bunch of clustering stuff that is using IDs, IPs, etc. | 17:20 |
| vladimir3p | vishy: this is another option we are exploring - just add more instances and add them into cluster | 17:20 |
| vladimir3p | (cluster running within instances) | 17:21 |
| kbringard | anyone know what role I need to assign a user in keystone to allow them to be able to associate floating IPs? | 17:21 |
| vishy | kbringard: using euca? | 17:21 |
| kbringard | oui | 17:21 |
| kbringard | I added the EC2 credentials | 17:21 |
| vishy | kbringard: netadmin | 17:21 |
| vishy | kbringard: sysadmin for launching instances | 17:21 |
| openstackgerrit | Ghe Rivero proposed a change to openstack/quantum: Added timeout flag to ovs-vsctl to avoid infinte waiting https://review.openstack.org/1715 | 17:21 |
| *** lorin1 has joined #openstack-dev | 17:22 | |
| vladimir3p | vishy: what do you think about the best option from Nova side? probably allow rebuild to receive a new host parameter? | 17:22 |
| vishy | vladimir3p: it seems like it would be way better to not depend on unique information like id | 17:22 |
| vishy | use name or metadata | 17:22 |
| vishy | vladimir3p: you can already specify ips when booting | 17:22 |
| kbringard | ok, right… but I'm kind of confused because keystone has a completely different role system… my user has "Member" for the appropriate tenant | 17:23 |
| vishy | vladimir3p: so if you can remove the dependency on unique data, then I don't think you need specific nova changes | 17:23 |
| *** jakedahn has quit IRC | 17:23 | |
| kbringard | and I've added the EC2 credentials | 17:23 |
| vishy | vladimir3p: unless you need to take data with you | 17:23 |
| *** jakedahn has joined #openstack-dev | 17:23 | |
| openstackgerrit | Brad Hall proposed a change to openstack/nova: Fix QuantumManager update_dhcp calls https://review.openstack.org/1939 | 17:23 |
| kbringard | I can do everything except associate floating IPs | 17:23 |
| vishy | kbringard: add a sysadmin and a netadmin role | 17:23 |
| vishy | kbringard: devstack does exactly that | 17:24 |
| kbringard | (launch instances, allocate floaters, etc) | 17:24 |
| kbringard | ah, OK | 17:24 |
| kbringard | fair enough :-) | 17:24 |
| kbringard | thanks | 17:24 |
| *** ayoung has joined #openstack-dev | 17:24 | |
| vladimir3p | vishy: yeah, what we are also doing is moving IP between instances for fail-over case. | 17:24 |
| vladimir3p | vishy: so we depend on resources associated with instance... | 17:25 |
| vladimir3p | vishy: of course, there are some workarounds | 17:25 |
| vishy | kbringard: https://github.com/openstack-dev/devstack/blob/master/files/keystone_data.sh#L17 | 17:25 |
| vishy | vladimir3p: so you can request an ip via api | 17:26 |
| kbringard | vishy: yoda man, thanks | 17:26 |
| vishy | vladimir3p: so it seems like you could just terminate the old vm and launch another one with the same ip | 17:26 |
| kbringard | my confusion was coming because I allocated via the dashboard, which uses the nova api | 17:26 |
| vishy | kbringard: yeah no rbac there yet | 17:27 |
| vladimir3p | vishy: ok, we need to explore the option to specify the IP during the instance creation | 17:27 |
| vishy | vladimir3p: I'm just trying to avoid adding specialized logic if possible | 17:27 |
| vladimir3p | vishy: I've not seen in in Diablo... I guess | 17:27 |
| vladimir3p | vishy: sure, me too :-) | 17:28 |
| vishy | vladimir3p: but i think we could generalize live/block_migrate resize/rebuild | 17:28 |
| vishy | and allow for a few extra parameters | 17:28 |
| vishy | if there are some resources you need to keep | 17:28 |
| vladimir3p | vishy: seems like it is more suitable for resize/rebuild | 17:28 |
| vishy | vladimir3p: I worry because there is a lot of duplicated code there | 17:28 |
| vishy | they all have to do a series of checks, then initialize a move | 17:28 |
| vishy | then verify | 17:29 |
| vishy | and rollback if there is an issue | 17:29 |
| vladimir3p | vishy: how about cleaning the cache (the use case when host is down and we rebuild instance on another host) | 17:29 |
| vishy | live/block is move every resource | 17:29 |
| vladimir3p | vishy: I mean cleaning the cache on failed node when it will re-appear | 17:29 |
| vishy | resize is move disk data only | 17:29 |
| openstackgerrit | Brad Hall proposed a change to openstack/nova: Add NAT/gateway support to QuantumManager https://review.openstack.org/1940 | 17:29 |
| vishy | sounds like you want move identity without data | 17:30 |
| vladimir3p | vishy: exactly | 17:30 |
| vishy | seems like we need a general move_instance | 17:30 |
| vishy | with various parameters | 17:30 |
| vishy | there is a feature-parity blueprint for making resize work with kvm | 17:30 |
| vladimir3p | we wanted to implement it for KVM, but decided that current functionality will not fit us (as is) | 17:31 |
| vladimir3p | so, seems like new generic move_instance will fit better | 17:31 |
| *** ayoung_ has joined #openstack-dev | 17:32 | |
| vladimir3p | ok, so we will explore the option to re-use IPs during instance spawning and if it will not work for us we will register a BP for generic move_instance... | 17:32 |
| *** ayoung has quit IRC | 17:33 | |
| vladimir3p | vishy: on a side node - how did it finish with stable/diablo package repositories? Is there a repo for keystone/nova/etc packages for stable/diablo? | 17:34 |
| vladimir3p | or should we build packages by our own? | 17:34 |
| *** ollie1 has left #openstack-dev | 17:34 | |
| openstackgerrit | Verification of a change to openstack/glance failed: Added some examples for "glance add" https://review.openstack.org/1888 | 17:39 |
| *** jdurgin has joined #openstack-dev | 17:42 | |
| *** negronjl_ has quit IRC | 17:42 | |
| *** code_franco has quit IRC | 17:43 | |
| vishy | vladimir3p: good deal | 17:45 |
| vishy | vladimir3p: there isn't | 17:45 |
| vishy | I'm bringing it up in the release meeting today though | 17:45 |
| vladimir3p | vishy: great. thanks | 17:45 |
| *** code_franco has joined #openstack-dev | 17:46 | |
| vladimir3p | vishy: imho it will be extremely useful. not sure if people would like to use "pure" diablo packages | 17:46 |
| vishy | i agree | 17:46 |
| mtaylor | vishy, vladimir3p: it's on my radar/todo list too. the current situation is certainly a problem | 17:46 |
| *** anotherjesse has joined #openstack-dev | 17:51 | |
| openstackgerrit | Brian Waldon proposed a change to openstack/nova: Convert compute manager delete methods to objects https://review.openstack.org/1928 | 18:01 |
| *** zns has quit IRC | 18:03 | |
| *** dprince has joined #openstack-dev | 18:08 | |
| openstackgerrit | Lorin Hochstein proposed a change to openstack/glance: Added some examples for "glance add" https://review.openstack.org/1888 | 18:11 |
| *** cp16net has joined #openstack-dev | 18:11 | |
| zul | mtaylor: hey do you know why the tarballs jenkins jobs are not running properly? | 18:15 |
| mtaylor | zul: looking | 18:16 |
| mtaylor | zul: which ones? | 18:18 |
| zul | mtaylor: dude all of them...horizon wasnt running until i did it manually | 18:18 |
| zul | glance had to be done manually on friday, same as nova | 18:19 |
| mtaylor | zul: well... horizon wasn't configured to be triggered by anything - that's now fixed | 18:20 |
| jaypipes | anybody seen any keystone devs today? | 18:22 |
| zul | they are suppose to be meeting today | 18:23 |
| *** openstackjenkins has quit IRC | 18:23 | |
| *** openstackjenkins has joined #openstack-dev | 18:24 | |
| *** adjohn has quit IRC | 18:25 | |
| *** vladimir3p has quit IRC | 18:25 | |
| jaypipes | zul: the keystone devs? | 18:30 |
| zul | jaypipes: yeah it just finished | 18:30 |
| jaypipes | mtaylor, zul, bcwaldon: so, I'd like to get this one done before the IRC meeting today: https://bugs.launchpad.net/glance/+bug/861650. Source code branch and packaging branch both need to go in together. After that, we'll need to upload new packages to PyPI... | 18:31 |
| uvirtbot | Launchpad bug 861650 in glance "Glance client packaging uses non-required dependencies" [Medium,In progress] | 18:31 |
| zul | jaypipes: i can merge it right now if you want | 18:33 |
| jaypipes | zul: well, that's the thing... we need to coordinate it all... | 18:34 |
| jaypipes | bcwaldon: care to review https://review.openstack.org/1619? | 18:34 |
| bcwaldon | yeah, on it | 18:34 |
| *** deshantm_laptop has joined #openstack-dev | 18:34 | |
| zul | jaypipes: ok lemme know when you need it | 18:34 |
| zul | ill stage it now | 18:35 |
| jaypipes | zul: cheers | 18:35 |
| *** jog0 has joined #openstack-dev | 18:35 | |
| *** darraghb has quit IRC | 18:35 | |
| *** openstackjenkins has quit IRC | 18:35 | |
| zul | jaypipes: im just a push away | 18:35 |
| *** openstackjenkins has joined #openstack-dev | 18:36 | |
| jaypipes | openstackjenkins: welcome. | 18:37 |
| openstackjenkins | jaypipes did you mean me? Unknown command 'welcome.' | 18:37 |
| openstackjenkins | Use '!jenkins help' to get help! | 18:37 |
| jaypipes | lol | 18:37 |
| mtaylor | curious about that ... | 18:38 |
| mtaylor | !jenkins help | 18:38 |
| openstackjenkins | Available commands: | 18:38 |
| openstackjenkins | abort <job> - specify which job to abort | 18:38 |
| openstack | mtaylor: Error: "jenkins" is not a valid command. | 18:38 |
| openstackjenkins | alias [<alias> [<command>]] - defines a new alias, deletes one or lists all existing aliases | 18:38 |
| openstackjenkins | botsnack [<snack>] - om nom nom | 18:38 |
| openstackjenkins | build <job> [now|<delay>[s|m|h]] [<parameterkey>=<value>]* - schedule a job build, with standard, custom or no quiet period | 18:38 |
| openstackjenkins | cb - list jobs which are currently in progress | 18:38 |
| openstackjenkins | comment <job> <build-#> <comment> - adds a description to a build | 18:38 |
| openstackjenkins | currentlyBuilding - list jobs which are currently in progress | 18:38 |
| openstackjenkins | h [<job>|-v <view>] - show the health of a specific job, jobs in a view or all jobs | 18:38 |
| openstackjenkins | health [<job>|-v <view>] - show the health of a specific job, jobs in a view or all jobs | 18:38 |
| openstackjenkins | jobs [<job>|-v <view>] - show the status of a specific job, jobs in a view or all jobs | 18:38 |
| openstackjenkins | q - show the state of the build queue | 18:38 |
| openstackjenkins | queue - show the state of the build queue | 18:38 |
| openstackjenkins | s [<job>|-v <view>] - show the status of a specific job, jobs in a view or all jobs | 18:38 |
| openstackjenkins | schedule <job> [now|<delay>[s|m|h]] [<parameterkey>=<value>]* - schedule a job build, with standard, custom or no quiet period | 18:38 |
| openstackjenkins | status [<job>|-v <view>] - show the status of a specific job, jobs in a view or all jobs | 18:38 |
| openstackjenkins | testresult [<job>|-v <view>] - show the test results of a specific job, jobs in a view or all jobs | 18:38 |
| openstackjenkins | userstat <username> - prints information about a Hudson user | 18:38 |
| mtaylor | well. | 18:38 |
| bcwaldon | mtaylor: feel free to merge this: https://code.launchpad.net/~jaypipes/glance/ubuntu/+merge/82318 | 18:44 |
| mtaylor | zul: ^^^ | 18:44 |
| *** zaitcev has quit IRC | 18:45 | |
| zul | done | 18:45 |
| anotherjesse | mr api (aka bcwaldon) - do you know if we expose information about the (minor) version of the service running in nova or glance? | 18:46 |
| mtaylor | zul: jenkins was generally unhappy. have re-started and all of the polling-based jobs are happy again | 18:46 |
| zul | mtaylor: ah cool...thanks | 18:47 |
| bcwaldon | anotherjesse: is 'GET /' what you want? | 18:47 |
| bcwaldon | anotherjesse: that gives a listing of minor versions (currently just 2.0) | 18:47 |
| bcwaldon | anotherjesse: glance has something similar | 18:47 |
| anotherjesse | bcwaldon: I'm thinking more info - like jenkin's does for /systeminfo (working on a screenshot for you) | 18:48 |
| anotherjesse | bcwaldon: http://stsh.me/wM | 18:49 |
| anotherjesse | for debugging it would be nice if we could expose that sort of thing either through a cli or api | 18:49 |
| bcwaldon | anotherjesse: I would love to have something that exposes more info, but right now it's literally just the minor version and a link to the docs | 18:50 |
| anotherjesse | bcwaldon: time to search to see if others have a blueprint or start drafting -- know if pub cloud has similar goals? | 18:50 |
| bcwaldon | anotherjesse: doubt it | 18:51 |
| *** GheRivero has left #openstack-dev | 18:51 | |
| anotherjesse | it would be nice if major services (glance, keystone, swift and nova) use the same output / api | 18:51 |
| bcwaldon | anotherjesse: do you mean the same endpoint? or just format? | 18:52 |
| anotherjesse | bcwaldon: if it could be: url from endpoint + /systemInfo (or something like that) | 18:52 |
| mtaylor | ++ | 18:52 |
| bcwaldon | anotherjesse: ah, okay. Yeah, that would be pretty useful. It may make sense to put that in an admin-only api | 18:53 |
| anotherjesse | bcwaldon: ya - was thinking that - if we need a subset that is exposed to users it seems like it should be part of the response to http://endpoint/ (like the api version) | 18:53 |
| anotherjesse | (eg a different issue) | 18:53 |
| bcwaldon | sure, I'm with you | 18:54 |
| openstackgerrit | Brian Waldon proposed a change to openstack/nova: Convert compute manager delete methods to objects https://review.openstack.org/1928 | 18:59 |
| *** dolphm has joined #openstack-dev | 19:00 | |
| *** negronjl has quit IRC | 19:01 | |
| openstackgerrit | Adrian Smith proposed a change to openstack/nova: Fix for bug 887712 https://review.openstack.org/1926 | 19:01 |
| uvirtbot | Launchpad bug 887712 in openstack-qa "instance_update with uuid as instance_Id and metadata fails" [Medium,In progress] https://launchpad.net/bugs/887712 | 19:01 |
| openstackgerrit | Alex Meade proposed a change to openstack/nova: reboot & rebuild to use uuids in compute manager https://review.openstack.org/1943 | 19:04 |
| *** dprince has quit IRC | 19:05 | |
| *** negronjl has joined #openstack-dev | 19:06 | |
| openstackgerrit | Naveed Massjouni proposed a change to openstack/nova: Removing line dos line endings in vmwareapi_conn.py https://review.openstack.org/1944 | 19:07 |
| *** vladimir3p has joined #openstack-dev | 19:09 | |
| *** bengrue has joined #openstack-dev | 19:10 | |
| anotherjesse | mtaylor: is jenkins merging stuck? usually gerrit gets updated almost instantly after the "approved" | 19:13 |
| anotherjesse | (nova) | 19:13 |
| mtaylor | anotherjesse: looking | 19:13 |
| mtaylor | anotherjesse: I don't see anything stuck - do you have a review you're looking at? | 19:14 |
| anotherjesse | review 1935 | 19:14 |
| mtaylor | ah. ok. see it. | 19:14 |
| mtaylor | investigating | 19:14 |
| anotherjesse | mtaylor: sorry to always be pinging with "hey this is borked" | 19:15 |
| anotherjesse | mtaylor: hope all is well :) | 19:15 |
| mtaylor | anotherjesse: nope. that's what I'm here for :) | 19:15 |
| mtaylor | anotherjesse: and yes - all is going swimmingly! (I'm going to have a bunch of blueprints posted for all of the CI-related stuff later today. should be fun reading) | 19:15 |
| *** scottjg has joined #openstack-dev | 19:16 | |
| openstackgerrit | Justin Shepherd proposed a change to openstack/keystone: Fixes bug lp:897819 https://review.openstack.org/1945 | 19:24 |
| *** openstackjenkins has quit IRC | 19:25 | |
| *** openstackjenkins has joined #openstack-dev | 19:26 | |
| *** pixelbeat has quit IRC | 19:34 | |
| *** novas0x2a|laptop has joined #openstack-dev | 19:34 | |
| rmk | Maybe someone here can speak to this. We're seeing behavior whereby if rabbitmq is restarted, all compute nodes also need to be restarted, otherwise new jobs will fail. | 19:43 |
| rmk | Are there any recommendations or workarounds to deal with this? | 19:43 |
| anotherjesse | rmk: which version of nova? | 19:45 |
| vishy | rmk: hmm, it seems like that has been fixed before | 19:45 |
| rmk | Latest stable/diablo. | 19:45 |
| vishy | rmk: although we may not have a means for reconnecting if a connection fails | 19:45 |
| vishy | comstud: thoughts on ^^ | 19:45 |
| anotherjesse | vishy: ya - was just noting that it would be a good ci test to add w.r.t. the chaos monkey | 19:45 |
| rmk | We're actually going through stability tests now to see what we need to address and this was one of the major issues we encountered | 19:46 |
| comstud | vishy: Other than 'it should work' with kombu, no ;) | 19:46 |
| rmk | Seems to be fine in regards to mysql -- it can be restarted issues | 19:47 |
| comstud | rmk: you using kombu or carrot? | 19:47 |
| vishy | mtaylor: what happened to the ci meeting? | 19:47 |
| mtaylor | vishy: timezone fail | 19:47 |
| anotherjesse | mtaylor: it wasn't me this time! | 19:47 |
| mtaylor | vishy: or daylight savings fail, I should say. all my fault | 19:47 |
| vishy | comstud: I know we have reconnection logic, but I don't know if it handles reconnecting an existing connection | 19:47 |
| rmk | comstud: kombu | 19:47 |
| vishy | comstud: I | 19:48 |
| rmk | comstud: What's the approrpiate method for verifying? All the log messages seem to indicate it's kombu. | 19:48 |
| vishy | comstud: think that maybe it just tries to reconnect if the initial connection fails | 19:48 |
| comstud | vishy: I added support for reestablishing queues and so forth as necessary | 19:48 |
| comstud | with kombu | 19:48 |
| vishy | rmk: what is the failure you get? | 19:48 |
| comstud | yeah, I'd like to see a traceback | 19:49 |
| rmk | I'm going to reproduce this and get you guys logs now | 19:49 |
| comstud | or other information | 19:49 |
| comstud | Thanks | 19:49 |
| vishy | rmk: cool | 19:49 |
| comstud | I've tested this before, but it's been a while | 19:49 |
| comstud | I had stuff going on while I restarted rabbit.. and it all reconnected and worked fine | 19:49 |
| bcwaldon | mtaylor: why are glance jobs failing in jenkins? | 19:49 |
| *** openstackjenkins has quit IRC | 19:49 | |
| *** Tribaal has quit IRC | 19:49 | |
| bcwaldon | mtaylor: ah, something going on behind the scenese | 19:50 |
| rmk | As a side note, it would be nice to have retries on both rabbit and mysql connections. | 19:50 |
| bcwaldon | mtaylor: please resume ;) | 19:50 |
| rmk | The nova flag for retries doesn't seem to do anything. | 19:50 |
| *** openstackjenkins has joined #openstack-dev | 19:50 | |
| mtaylor | bcwaldon: operator error on my fault. I upgraded a plugin which accidentally caused the git plugin to be upgraded, but that borked the gerrit trigger plugin version we're using ... downgrading the git plugin right now | 19:50 |
| bcwaldon | mtaylor: don't let it happen again! | 19:51 |
| mtaylor | bcwaldon: jesus. I would really like to avoid it strongly :) | 19:51 |
| bcwaldon | mtaylor: sure... | 19:51 |
| openstackgerrit | Tyler Smith proposed a change to openstack/quantum: Fixing find_config_file after packaging changes https://review.openstack.org/1946 | 19:51 |
| mtaylor | bcwaldon: that's a lie. secretly I like breaking things from time to time so that you'll all like me when I fix them again | 19:51 |
| mtaylor | bcwaldon: it's my background as a consultant coming out ;) | 19:52 |
| bcwaldon | mtaylor: well you shouldnt admit to breaking them | 19:52 |
| bcwaldon | mtaylor: blame it on jeblair and I'll like you even more | 19:52 |
| mtaylor | bcwaldon: hrm. that's a godo point | 19:52 |
| rmk | comstud: Would it improve reliability to use persistent queues? | 19:52 |
| mtaylor | bcwaldon: no one would ever believe me if I blamed anything on jeblair | 19:52 |
| bcwaldon | mtaylor: true | 19:52 |
| mtaylor | bcwaldon: blaming soren on the other hand I think is fair game ... | 19:52 |
| *** yogirackspace has joined #openstack-dev | 19:53 | |
| comstud | rmk: Not sure where we should use them, really.. but it shouldn't matter much as far as this reconnecting issue | 19:53 |
| mtaylor | bcwaldon: any chance you have insane amounts of virtualenv zen? | 19:53 |
| bcwaldon | mtaylor: I actually just learned how to use it effectively...so yes? | 19:54 |
| bcwaldon | mtaylor: but probably no | 19:54 |
| mtaylor | bcwaldon: :) | 19:54 |
| * mtaylor is looking for someone to help him figure out why the nova-docs job is broken | 19:54 | |
| mtaylor | specifically - why I can't do build_sphinx inside of a venv | 19:54 |
| *** kimscheibel_ has joined #openstack-dev | 19:54 | |
| bcwaldon | mtaylor: can you share the output? | 19:55 |
| mtaylor | and why the error message says that a file isn't there - when the file most certainly IS there | 19:55 |
| mtaylor | bcwaldon: one sec | 19:55 |
| *** bencherian has joined #openstack-dev | 19:56 | |
| mtaylor | bcwaldon: hrm. something is weird... lemme get back to you on that | 19:56 |
| bcwaldon | mtaylor: kk, I probably couldn't solve the problem anyways | 19:56 |
| openstackgerrit | Verification of a change to openstack/glance failed: Fixes LP Bug#861650 - Glance client deps https://review.openstack.org/1619 | 19:56 |
| *** novas0x2a|laptop has quit IRC | 19:57 | |
| mtaylor | bcwaldon: ok. so that ^^^ looks like it might be an actual valid error? | 19:58 |
| bcwaldon | it was a pretty invasive change, wouldnt be surprised | 19:58 |
| bcwaldon | looking now | 19:58 |
| mtaylor | bcwaldon: does the output of that job look like it's trying to merge that change into stable/diablo? | 19:59 |
| *** joesavak has joined #openstack-dev | 19:59 | |
| mtaylor | bcwaldon: never mind. I can't read | 19:59 |
| soren | mtaylor: Blaming me for what? | 20:00 |
| bcwaldon | mtaylor: yeah, its a legitimate merge conflict with master | 20:00 |
| bcwaldon | mtaylor: jaypipes needs to fix that | 20:00 |
| bcwaldon | jaypipes: but he's in a PPB meeting now | 20:00 |
| bcwaldon | mtaylor: whoops ^ | 20:00 |
| bcwaldon | mtaylor: I'll just fix it for him | 20:00 |
| *** GheRivero has joined #openstack-dev | 20:00 | |
| *** kimscheibel_ has left #openstack-dev | 20:02 | |
| bcwaldon | mtaylor: actually, I need to talk to him about a conflict here | 20:02 |
| bcwaldon | mtaylor: we'll have to wait | 20:02 |
| openstackjenkins | Project nova-docs build #1,684: FAILURE in 1 min 56 sec: https://jenkins.openstack.org/job/nova-docs/1684/ | 20:02 |
| openstackjenkins | * Rick Harris: Adding downgrade for migration 57 plus test | 20:02 |
| openstackjenkins | * Rick Harris: Bind engine to metadata in migration 054. | 20:02 |
| openstackjenkins | * Rick Harris: Defining volumes table to allow FK constraint. | 20:02 |
| openstackjenkins | * mark.washenberger: Add logging, error handling to the xenstore lib. | 20:02 |
| openstackjenkins | * Vishvananda Ishaya: Separate metadata api into its own service | 20:02 |
| openstackjenkins | * lorin: Reference Ron Pedde's cleanup script for DevStack. | 20:02 |
| openstackjenkins | * Alvaro: Fixes bug 890206 | 20:02 |
| openstackjenkins | * brian.lamar: Remove hostname update() logic in compute.API | 20:02 |
| uvirtbot | Launchpad bug 890206 in nova "Unable to attach volume to a xen PVM" [Undecided,Fix committed] https://launchpad.net/bugs/890206 | 20:02 |
| openstackjenkins | * brian.waldon: Convert security_group calls to use instance objs | 20:02 |
| openstackjenkins | * Vishvananda Ishaya: Makes sure gateways forward properly | 20:02 |
| openstackjenkins | * Rick Harris: Replacing disk config extension to match spec. | 20:02 |
| openstackjenkins | * Vishvananda Ishaya: Make run instances respect availability zone | 20:02 |
| openstackjenkins | * brian.waldon: Convert remaining calls to use instance objects | 20:02 |
| openstackjenkins | * brian.lamar: Refactoring/cleanup of some view builders. | 20:02 |
| openstackjenkins | * lorin: Added RST docs on how to use gettext. | 20:02 |
| openstackjenkins | * Alvaro: Add a "libvirt_disk_prefix" flag to libvirt driver | 20:02 |
| openstackjenkins | * brian.waldon: Creating new v2 namespace in nova.api.openstack | 20:02 |
| openstackjenkins | * lorin: Call df with -k instead of -B1 | 20:03 |
| openstackjenkins | * lorin: Updated development environment docs. | 20:03 |
| openstackjenkins | * alex.meade: rescue/unrescue in compute manager to use uuids | 20:03 |
| openstackjenkins | * naveedm9: Updating set_admin_password in compute.api and compute.manager to use instance | 20:03 |
| openstackjenkins | * alex.meade: reset/inject network info in compute to use uuid | 20:03 |
| openstackjenkins | * Brad Hall: Fix for bug 894431 | 20:03 |
| openstackjenkins | * lorin: Fix deprecation warnings | 20:03 |
| uvirtbot | Launchpad bug 894431 in nova "linux_net ovsinterfacedriver is setting the wrong iface-id" [Undecided,Fix committed] https://launchpad.net/bugs/894431 | 20:03 |
| openstackjenkins | * ewan.mellor: Fix trivial fourth quote in docstring. | 20:03 |
| openstackjenkins | * lorin: New docs: unit tests, Launchpad, Gerrit, Jenkins | 20:03 |
| openstackjenkins | * ewan.mellor: Refactor a few things inside the xenapi unit tests. | 20:03 |
| openstackjenkins | * ewan.mellor: Bug #897054: stack crashes with AttributeError on e.reason if the server returns an error | 20:03 |
| openstackjenkins | * ewan.mellor: Bug #897091: "nova actions" fails with HTTP 400 / TypeError if a server action has been performed | 20:03 |
| uvirtbot | Launchpad bug 897054 in nova "stack crashes with AttributeError on e.reason if the server returns an error" [Undecided,Fix committed] https://launchpad.net/bugs/897054 | 20:03 |
| soren | Argh! | 20:03 |
| openstackjenkins | * likitha.shetty: Fix Bug #891718 | 20:03 |
| openstackjenkins | * oliver.leahy: Fixes bug 888649 | 20:03 |
| uvirtbot | Launchpad bug 897091 in nova ""nova actions" fails with HTTP 400 / TypeError if a server action has been performed" [Medium,Fix committed] https://launchpad.net/bugs/897091 | 20:03 |
| openstackjenkins | * markw: Use instance uuids for consoles and diagnostics. | 20:03 |
| openstackjenkins | * naveedm9: Updating {add,remove}_fixed_ip_from_instance in compute.api and compute.manager | 20:03 |
| openstackjenkins | * mark.washenberger: start/stop in compute manager to use uuids | 20:03 |
| uvirtbot | Launchpad bug 891718 in nova "nova components always log to standard error, even when syslog is turned on" [Wishlist,Fix committed] https://launchpad.net/bugs/891718 | 20:03 |
| openstackjenkins | * Kevin L. Mitchell: Templatize extension handling. | 20:03 |
| uvirtbot | Launchpad bug 888649 in nova "Snapshots left in undeletable state" [Undecided,Fix committed] https://launchpad.net/bugs/888649 | 20:03 |
| mtaylor | jesus | 20:03 |
| *** zaitcev has joined #openstack-dev | 20:04 | |
| *** johnpur has joined #openstack-dev | 20:04 | |
| *** ChanServ sets mode: +v johnpur | 20:04 | |
| bcwaldon | look what you did | 20:04 |
| mtaylor | bcwaldon: https://jenkins.openstack.org/job/nova-docs/1684/console | 20:05 |
| bcwaldon | mtaylor: ha, yeah I have no idea | 20:06 |
| *** adiantum has quit IRC | 20:06 | |
| lorin1 | mtaylor: That nova-docs thing was my bad: https://bugs.launchpad.net/nova/+bug/888102 | 20:06 |
| uvirtbot | Launchpad bug 888102 in nova "In RST docs, some module names now missing a dot" [Low,Confirmed] | 20:06 |
| mtaylor | lorin1: no... there's a more fundamental issue in the jenkins builder | 20:07 |
| mtaylor | lorin1: BUT - I'm happy to let you take blame for a little bit | 20:07 |
| lorin1 | mtaylor: Ah, I see it, there are I/O errors going on there. | 20:07 |
| mtaylor | yeah. and a file it can't find that's actually there | 20:08 |
| *** dprince has joined #openstack-dev | 20:10 | |
| openstackgerrit | Monty Taylor proposed a change to openstack/openstack-ci: Added code to inject source project name tags. https://review.openstack.org/1947 | 20:14 |
| *** koolhead17 has left #openstack-dev | 20:16 | |
| openstackjenkins | Project nova-docs build #1,685: STILL FAILING in 1 min 19 sec: https://jenkins.openstack.org/job/nova-docs/1685/ | 20:16 |
| openstackjenkins | Jesse Andrews: removed logic of throwing exception if no floating ip | 20:16 |
| zaitcev | Guys, what if I wanted to add a printf() to some middleware, what would be the best way? I mean, aside from fp=open("/tmp/dump","a") print >>fp,response.body | 20:17 |
| zaitcev | (I'm looking at reasons why legacy_token_auth.py fails in Keystone) | 20:18 |
| *** jakedahn has quit IRC | 20:18 | |
| rmk | comstud: With durable queues enabled in nova, are messages also durable? | 20:18 |
| openstackgerrit | Dolph Mathews proposed a change to openstack/keystone: Added documentation for SQL tables https://review.openstack.org/1948 | 20:24 |
| vishy | zaitcev: if it is middleware, you should be able to use logging | 20:24 |
| openstackgerrit | Verification of a change to openstack/openstack-manuals failed: Changed keystone port from 5001 to 35357 https://review.openstack.org/1858 | 20:24 |
| zaitcev | vishy: What class or function is it? find . -name '*log*' finds nothing. I just need a hint, to pull out the right page. | 20:26 |
| vishy | zaitcev: is this in keystone? | 20:27 |
| zaitcev | vishy: Yes | 20:27 |
| vishy | zaitcev: and the middleware runs in which binary? | 20:27 |
| vishy | zaitcev: sorry, I'm not so sure how logging is set up in keystone. dolphm ? ^^ | 20:28 |
| *** Kiall has quit IRC | 20:28 | |
| zaitcev | The middleware is inserted into a WSGI stack it seems, using the pipeline specification in keystone.conf. I am unclear on the details. It is inside the python interpterer running some kind of webserver, initiated by ./bin/keystone. It quickly gets into server.start() and thereabout so I'm quite lost. | 20:29 |
| *** Kiall has joined #openstack-dev | 20:31 | |
| openstackgerrit | Brian Waldon proposed a change to openstack/nova: Convert compute manager delete methods to objects https://review.openstack.org/1928 | 20:31 |
| *** jsavak has joined #openstack-dev | 20:31 | |
| openstackjenkins | Project nova-docs build #1,686: STILL FAILING in 1 min 36 sec: https://jenkins.openstack.org/job/nova-docs/1686/ | 20:32 |
| openstackjenkins | * Chris Behrens: Fix RPC responses to allow None response correctly. | 20:32 |
| openstackjenkins | * adrian_f_smith: Fix for bug 887712 | 20:32 |
| uvirtbot | Launchpad bug 887712 in openstack-qa "instance_update with uuid as instance_Id and metadata fails" [Medium,In progress] https://launchpad.net/bugs/887712 | 20:32 |
| * mtaylor punches openstackjenkins in the face | 20:32 | |
| zaitcev | I am somewhat familiar with the concept of the middleware now. I worked with mod_wsgi and put together a small webapp plugging into WSGI. Back at the Bexar meeting, I was at creiht's preso about S3 and he was throwing the m-word in every other sentence. I had _no clue_ what he was talking about. | 20:33 |
| zaitcev | No idea about the implications: what configs middleware can consume, what status to keep between requests etc | 20:34 |
| *** joesavak has quit IRC | 20:34 | |
| openstackgerrit | Verification of a change to openstack/keystone failed: Check that endpointTemplate ID is valid in endpoint add cmd (#897749) https://review.openstack.org/1941 | 20:37 |
| vishy | mtaylor: is there any way to set a blueprint or bug after the fact if someone neglects to link properly in the commit message? | 20:39 |
| markmc | zaitcev, might be worth looking at PasteDeploy docs, that'll explain the config side of things anyway | 20:39 |
| jaypipes | jeblair: ping | 20:40 |
| markmc | zaitcev, for logging, should be able to simply do import logging and logging.debug("....") | 20:40 |
| * markmc tries that out | 20:40 | |
| mtaylor | vishy: to set it where? to set up the link from the bug to the change? or the other way around? | 20:40 |
| vishy | either | 20:40 |
| vishy | so in launchpad if someone forgot to link, I could go in and link it for them | 20:40 |
| vishy | now i have to ask them to change their commit message | 20:41 |
| vishy | is there a secret admin way around? | 20:41 |
| mtaylor | vishy: yeah - all gerrit is doing on launchpad is adding a comment with a link - so you could do that as well | 20:41 |
| vishy | mtaylor: and changing the topic on launchpad? | 20:41 |
| rmk | There are quite a few major scalability issues with the Dashboard | 20:41 |
| rmk | Not sure how high those are on the priority list | 20:42 |
| rmk | We're seeing them in our testing here | 20:42 |
| mtaylor | vishy: not sure what you mean by that? | 20:42 |
| zaitcev | markmc: I'll try that. I think I had some examples on use of generic Python logging. The thing is, if I look at keystone/common/wsgy.py, I find that we create a log (using WritableLogger()), pass it down to eventlet and that is what points to normal log file -- magically. | 20:42 |
| vishy | set the topic on the gerrit side | 20:43 |
| zykes- | rmk: which ones ? | 20:43 |
| openstackgerrit | Jay Pipes proposed a change to openstack/keystone: Fixes a number of configuration/startup bugs https://review.openstack.org/1936 | 20:43 |
| *** johnpur has quit IRC | 20:43 | |
| *** viraptor has quit IRC | 20:43 | |
| rmk | zykes-: Related to the number of instances currently running, or have run over the course of a given month. | 20:43 |
| mtaylor | vishy: ah. I don't believe there is a way to change the topic on the gerrit side ... however, perhaps jeblair knows something I don't | 20:43 |
| vishy | mtaylor: reviewday.ohthree.com keys off of the topics | 20:43 |
| rmk | zykes-: Both have scale issues. | 20:43 |
| zaitcev | markmc: So I highly doubt that logging.debug() would go where it should... Probably would end in stderr. I'll try it when I get near my big box with Openstack setup. | 20:43 |
| rmk | Fire up 100 VMs and load any instances page which would display them and watch how long it takes to load. | 20:43 |
| rmk | In the process it slams keystone/nova-api. | 20:43 |
| markmc | zaitcev, WritableLogger, interesting | 20:44 |
| *** coli has quit IRC | 20:44 | |
| rmk | Or, start/terminate ~1000 instances, leaving none running and watch how long the overview page takes to load, slamming nova-api in the process. | 20:44 |
| mtaylor | vishy: gotcha. not sure. | 20:44 |
| mtaylor | also, I'd never seen that tool before. | 20:44 |
| rmk | Seems like it's the compute/hr calculation in that case causing the delay and associated load. | 20:45 |
| rmk | Either way, these aren't huge numbers in the "cloud" context | 20:45 |
| markmc | zaitcev, why are you specifically interested in legacy_auth? | 20:45 |
| markmc | zaitcev, is the swift middleware using that? | 20:45 |
| vishy | mtaylor: pretty fancy huh? | 20:46 |
| vishy | mtaylor: there was an email on the list a few days ago about it | 20:46 |
| zaitcev | markmc: Good question... Docs are contradictory. Some say "supply /1.0" to swift -A, which would make it using legacy. Other docs mention /2.0, which is the native keystone with /2.0/token and the like, but I think that does not work | 20:46 |
| *** crobinso has quit IRC | 20:46 | |
| openstackgerrit | anotherjesse proposed a change to openstack-dev/devstack: these deps should come via git_clone https://review.openstack.org/1949 | 20:47 |
| markmc | zaitcev, ah, ok | 20:47 |
| openstackgerrit | Naveed Massjouni proposed a change to openstack/nova: Fixing get_info method implementations in virt drivers to accept instance_name instead of instance_id. The abstract class virt.ComputeDriver defines get_info as: def get_info(self, instance_name). blueprint: internal-uuids https://review.openstack.org/1950 | 20:47 |
| mtaylor | vishy: neat. | 20:47 |
| openstackgerrit | A change was merged to openstack/nova: Use uuids for compute manager agent update. https://review.openstack.org/1942 | 20:48 |
| markmc | zaitcev, try just running keystone-auth from the command line and you should see stdout - e.g. "Starting the Legacy Authentication component" | 20:49 |
| zaitcev | markmc: If I give swift -A http://localhost:5000/v1.0 it attempts to do the right thing and recognizes the X-Auth-Token. But it crashes because there's no service URL in the reply (X-Storage-Url in my case) | 20:49 |
| rmk | Had to wipe out the instances table to get performance to be acceptable again. | 20:50 |
| comstud | rmk: What do you mean by 'durable message'? | 20:50 |
| rmk | comstud: As in, a message placed into a queue also persists a restart. | 20:50 |
| jeblair | jaypipes: pong | 20:50 |
| comstud | rmk: I think that's just a function of making the queue durable | 20:50 |
| rmk | comstud: According the rabbit docs that's not the case. | 20:51 |
| comstud | rmk: Hm.. I know of nothing with AMQP about marking _messages_ durable | 20:51 |
| markmc | zaitcev, is the user registered under a tenant? I'm not sure if there's a concept of endpoints for unscoped auth tokens | 20:51 |
| markmc | zaitcev, could be completely wrong on that one | 20:51 |
| comstud | rmk: But I'm not an expert on this stuff | 20:51 |
| rmk | comstud: I'll see if I can find you a direct reference on this -- my peer here was the one who looked into it | 20:52 |
| zaitcev | markmc: I ran keystone-manage role grant Admin admin admin. The output of keystone-manage role list is not very helpful, it only lists roles, but not who has them. Same for user and tenant lists. | 20:53 |
| zaitcev | I figured the best way is to look at the source and guess what Keystone people meant to do here. | 20:53 |
| rmk | zaitcev: Unfortunately true for a lot of keystone's output right now. I usually end up going to the db. | 20:53 |
| jaypipes | jeblair: Hi Jim, hoping you could shed some light on this: http://paste.openstack.org/show/3551/ | 20:54 |
| jaypipes | jeblair: basically, I need to resolve a merge conflict in this branch: https://review.openstack.org/#change,1619. However, doing a git review -d XXXX is doing an automatic merge (instead of a rebase AFAICT) with origin/master and so it's not letting me amend the original commit. | 20:55 |
| markmc | zaitcev, well, in this example I don't get a serviceCatalog in the auth response if I don't pass tenantName - http://fpaste.org/vSrc/ | 20:56 |
| mtaylor | jaypipes: hrm. that might be something to do with how git review -d is working ... | 20:56 |
| markmc | zaitcev, interesting, legacy_auth has no concept of tenant | 20:57 |
| mtaylor | jaypipes: lemme take a look ... | 20:57 |
| jaypipes | mtaylor: appreciate it. thx :) | 20:57 |
| jeblair | jaypipes: what's the starting state of your tree? | 20:57 |
| zaitcev | markmc: fascinating, thanks | 20:58 |
| jaypipes | jeblair: you mean what branch/rev am I on when I do a git review -d XXXX? | 20:58 |
| jeblair | yep | 20:58 |
| ttx | Meeting starts in 2 min. in #openstack-meeting, please join | 20:58 |
| jaypipes | jeblair: origin/master latest rev | 20:58 |
| mtaylor | jaypipes: ok. doing that for me gets me into a branch called review/jay_pipes/bug/861650 | 20:59 |
| markmc | zaitcev, oh, I appear to get the service catalog just fine on master with that same example | 20:59 |
| mtaylor | jaypipes: with the commit 7a65a099f8092ae88f16ea8092ae35808627028c as the tip commit | 21:00 |
| markmc | zaitcev, I'll double check that | 21:00 |
| markmc | dolphm, does this ring any bells? No serviceCatalog from tenantName-less auth on stable/diablo, but works on master? | 21:00 |
| mtaylor | at that point I should just be able to edit something, do "git commit -a --amend ; git review" and it _should_ resubmit it | 21:00 |
| jeblair | mtaylor: that's not what happened for mo | 21:00 |
| jeblair | me | 21:00 |
| mtaylor | jeblair: oh yeah? GREAT | 21:01 |
| zaitcev | Oh, that master. OK, I'll rebuild from git. I was using 1.4.3 RPMs that apevec built. | 21:01 |
| jeblair | i ended up in the middle of a merge, like the pastebin, with 6bf61e8ce1882ea0f448aaffb7de8a826f19ea6e as the tip | 21:01 |
| markmc | zaitcev, keystone master, I mean | 21:01 |
| zaitcev | markmc: I'll continue when I get back home, thanks a lot for your help | 21:02 |
| markmc | zaitcev, np | 21:02 |
| mtaylor | jeblair: any chance we have git-review version skew (and/or a bug in our stale version detection code?) | 21:02 |
| openstackjenkins | Project nova-docs build #1,687: STILL FAILING in 1 min 30 sec: https://jenkins.openstack.org/job/nova-docs/1687/ | 21:02 |
| openstackjenkins | * mark.washenberger: Use uuids for compute manager agent update. | 21:02 |
| openstackjenkins | * alex.meade: reboot & rebuild to use uuids in compute manager | 21:02 |
| mtaylor | jeblair: I'm on 1.6 | 21:02 |
| jeblair | i'm on 1.7 | 21:02 |
| jeblair | and i updated to master like jay suggestied | 21:03 |
| jeblair | so i think one potential problem is that by doing a git pull of the change, we create a merge commit | 21:03 |
| jaypipes | right... needs to be a rebase. | 21:04 |
| mtaylor | _should_ be a fetch + checkout really | 21:04 |
| jeblair | which is fine if everything works, but if not, how does one get a modified change back to gerrit without pushing a merge commit. | 21:04 |
| mtaylor | but yes. you are right - it should certainly not be a pull | 21:05 |
| * mtaylor is still confused as to why it worked for him and not for jeblair/jaypipes - but it's clearly a bug | 21:05 | |
| *** zaitcev has quit IRC | 21:06 | |
| jeblair | did you start from the latest master? | 21:06 |
| mtaylor | I did | 21:06 |
| *** adiantum has joined #openstack-dev | 21:06 | |
| jeblair | git fetch https://review.openstack.org/p/openstack/glance refs/changes/19/1619/3 && git checkout FETCH_HEAD | 21:07 |
| jeblair | works for me | 21:07 |
| mtaylor | yes. I believe that's the right way to do it | 21:07 |
| *** mwhooker has joined #openstack-dev | 21:08 | |
| jeblair | jaypipes: so you can use that command instead of 'git review -d' to fix your immediate problem | 21:09 |
| jeblair | mtaylor: i'll file a bug | 21:09 |
| mtaylor | jeblair: thanks. I'm working on a patch right now | 21:09 |
| *** dprince has quit IRC | 21:10 | |
| jeblair | mtaylor: bug 897871 | 21:12 |
| uvirtbot | Launchpad bug 897871 in git-review "downloading a change doesn't work if there is a merge conflict" [Medium,Confirmed] https://launchpad.net/bugs/897871 | 21:12 |
| jaypipes | jeblair: yep, works. FYI, git checkout FETCH_HEAD -b XXXX is a little better, to prevent working in detached head state | 21:12 |
| jeblair | yeah, and consistent with what git-review is currently doing. | 21:13 |
| mtaylor | yes. that is how I will be fixing it in git-review | 21:14 |
| openstackgerrit | Monty Taylor proposed a change to openstack-ci/git-review: Use fetch/checkout instead of pull. https://review.openstack.org/1951 | 21:15 |
| *** troytoman-away is now known as troytoman | 21:15 | |
| mtaylor | jeblair: ^^ | 21:15 |
| jeblair | mtaylor: reviewing | 21:16 |
| mtaylor | jeblair: cool. I tested by using it to download jay's change | 21:17 |
| openstackjenkins | Project nova-docs build #1,688: STILL FAILING in 1 min 17 sec: https://jenkins.openstack.org/job/nova-docs/1688/ | 21:17 |
| openstackjenkins | * mark.washenberger: Use uuids for file injection. | 21:17 |
| openstackjenkins | * naveedm9: Removing line dos line endings in vmwareapi_conn.py | 21:17 |
| jeblair | i'm going to do that too. | 21:17 |
| * mtaylor has found the nova-docs problem - searching for a good solution now | 21:17 | |
| Kiall | Heya, I've noticed some dashboard issues with the latest python-novaclient.. Just wondering if these should be caught and ignored? "Permission denied: '/var/www/.novaclient_cached_keypair_uuids'" | 21:19 |
| jk0 | Kiall: would you mind filing a bug for that? | 21:21 |
| mtaylor | jeblair: I have found the worlds most disgusting workaround for the nova-docs issue ... but it works for now, and the real solution involves sorting out an interaction between sphinx, docutils and virtualenv | 21:21 |
| Kiall | jk0: sure, if failure to write is meant to be ignored, i can fix while I'm at it? | 21:21 |
| jk0 | sure, absolutely | 21:22 |
| jk0 | patches always welcome :) | 21:22 |
| *** adiantum has quit IRC | 21:22 | |
| jeblair | mtaylor: i'm not familiar with the nova-docs issue | 21:22 |
| *** dprince has joined #openstack-dev | 21:22 | |
| mtaylor | jeblair: the job has been failing since we moved to venv builders | 21:22 |
| *** novas0x2a|laptop has joined #openstack-dev | 21:23 | |
| openstackjenkins | Yippie, build fixed! | 21:24 |
| openstackjenkins | Project nova-docs build #1,689: FIXED in 4 min 14 sec: https://jenkins.openstack.org/job/nova-docs/1689/ | 21:24 |
| annegentle | double Yippie on that :) | 21:24 |
| openstackgerrit | Brian Waldon proposed a change to openstack/nova: Convert compute manager delete methods to objects https://review.openstack.org/1928 | 21:26 |
| mtaylor | notmyname: is there a general way to set default command line flags in my git config? | 21:27 |
| mtaylor | notmyname: for instance, if I wanted pull to default to --ff-only, or git tag to default to having -s? | 21:27 |
| markmc | mtaylor, merge.ff looks like the option you want | 21:29 |
| markmc | mtaylor, git config --global merge.ff true | 21:29 |
| mtaylor | markmc: sweet. can I ask where I would go to discover such a thing? | 21:30 |
| markmc | mtaylor, rtfm! | 21:30 |
| markmc | mtaylor, man git-config :) | 21:30 |
| mtaylor | thanks! | 21:31 |
| markmc | np :) | 21:31 |
| mtaylor | markmc: (I couldn't find the right FM to R) | 21:31 |
| *** negronjl has quit IRC | 21:31 | |
| *** lorin1 has quit IRC | 21:31 | |
| notmyname | mtaylor: ya, use git alias (I think) | 21:33 |
| notmyname | mtaylor: http://paste.openstack.org/show/3552/ | 21:33 |
| notmyname | mtaylor: ya, alias will work. see my ~/.gitconfig above | 21:34 |
| markmc | dolphm, any idea what happend here? https://jenkins.openstack.org/job/gate-keystone-unittests/55/console | 21:35 |
| markmc | dolphm, looks like a test timed out? don't think it's related to my change, and the tests pass locally | 21:35 |
| markmc | notmyname, oh, nice | 21:35 |
| *** rods has joined #openstack-dev | 21:38 | |
| *** negronjl has joined #openstack-dev | 21:38 | |
| dolphm | markmc: looking.. | 21:40 |
| *** dprince has quit IRC | 21:41 | |
| dolphm | markmc: that looks successful... wrong job id? | 21:41 |
| markmc | dolphm, bah :) https://jenkins.openstack.org/job/gate-keystone-unittests/56/ | 21:41 |
| dolphm | ah, yeah the management API is totally untested... shouldn't have affected 'unit tests' | 21:42 |
| jeblair | markmc: Build timed out (after 5 minutes). Marking the build as failed. | 21:43 |
| markmc | dolphm, well, interestingly just loading sampledata exercises it a little | 21:43 |
| jeblair | sadly that kind of gets lost with all the "....."'s | 21:43 |
| markmc | dolphm, it caught a buglet of mine in that patch :) | 21:43 |
| markmc | jeblair, yeah, saw the timeout alright | 21:43 |
| dolphm | markmc: yeah, but the results aren't exactly asserted lol | 21:43 |
| dolphm | jeblair: so, our tests are just taking > 5 minutes now? | 21:44 |
| dolphm | we did add an additional test configuration recently (ssl) which runs in series with the rest.. so that probably pushed it close to the edge | 21:45 |
| jeblair | it does look like sucessful runs are > 5 mins. shall i update the timout to 10 or 15 mins? | 21:45 |
| dolphm | jeblair: that's fine with me | 21:46 |
| dolphm | jeblair: jenkins is probably the only one that runs the entire test suite with every supported configuration, so it's not exactly indicative of a long feedback loop for developers anyway :P | 21:47 |
| *** dprince has joined #openstack-dev | 21:48 | |
| wwkeyboard | Is anyone else seeing launchpad take hours to process mails to lists.launchpad.net? | 21:54 |
| openstackgerrit | Kiall Mac Innes proposed a change to openstack/python-novaclient: Gracefully handle failure to cache UUID's. Bug #897885. https://review.openstack.org/1952 | 21:55 |
| uvirtbot | Launchpad bug 897885 in python-novaclient "When caching of UUID fails, the command should not fail." [Undecided,New] https://launchpad.net/bugs/897885 | 21:55 |
| wwkeyboard | I know Vek has seen this problem before, but it seems to be pretty bad today. | 21:55 |
| openstackgerrit | Kiall Mac Innes proposed a change to openstack/python-novaclient: Gracefully handle failure to cache UUID's. Bug #897885. https://review.openstack.org/1952 | 21:56 |
| uvirtbot | Launchpad bug 897885 in python-novaclient "When caching of UUID fails, the command should not fail." [Undecided,New] https://launchpad.net/bugs/897885 | 21:56 |
| Kiall | jk0: that gerrit review should fix the issue.. | 21:56 |
| Vek | as a point of information, I actually just had the test suite hang on me, running it on my dev box. No clue why. | 21:56 |
| dolphm | jeblair: can you retrigger that job whenever you increase the timout threshold? | 22:00 |
| *** rkukura has quit IRC | 22:01 | |
| jeblair | dolphm: done | 22:03 |
| *** jsavak has quit IRC | 22:04 | |
| *** dprince has quit IRC | 22:07 | |
| mtaylor | Kiall: as may be evidenced, I _would_ like to talk to you about involving you in some of the packaging and infrastructure work from the project level | 22:08 |
| *** mikeyp has joined #openstack-dev | 22:08 | |
| ttx | mtaylor: my point is that so far we have been focused on a release. Focusing on multiple parallel threads won't get better results | 22:08 |
| mtaylor | Kiall: of course, the conversation aroudn what _should_ be done is, well, ongoing :) | 22:08 |
| ttx | we should let the distributions do the work at which they excel | 22:08 |
| ttx | and stay out of it | 22:08 |
| mtaylor | ttx: I hear that - I'm just saying that our insistence on that approach isn't working | 22:09 |
| ttx | it's ok to produce test/eval packages | 22:09 |
| dolphm | jeblair: thanks! | 22:09 |
| markmc | ttx, can I ask again, do you think 2011.3.x releases or a bad idea ? | 22:09 |
| mtaylor | and I do not think that the distros are handling the use cases of our users | 22:09 |
| mtaylor | at all | 22:09 |
| ttx | it's not ok to produce "almost prod-ready" packages | 22:09 |
| mtaylor | I think the distros are handling the use cases of the distros | 22:09 |
| markmc | ttx, because IMHO, it helps downstreams do exactly as you say | 22:09 |
| dolphm | markmc: https://review.openstack.org/#change,1941 | 22:09 |
| ttx | markmc: I think they are a bad idea, but for a different reason | 22:10 |
| markmc | dolphm, cool, thanks | 22:10 |
| ttx | markmc: basically we have a process that ensures a certain level of quality for the dev release. And a completely different process for 2011.3.x | 22:10 |
| ttx | that makes me a sad bunny, when you can't be sure that 2011.3.x actually ends up being better | 22:11 |
| *** mnewby has joined #openstack-dev | 22:11 | |
| Kiall | mtaylor: sorry, was AFK | 22:11 |
| markmc | ttx, you think the stable branch process does less to ensure quality than the dev release process? | 22:12 |
| ttx | we could apply the same process, but that takes loads of efforts. Hence the dilution and loss of focus | 22:12 |
| Daviey | hmmmm | 22:12 |
| ttx | markmc: I'm pretty sure of it. | 22:12 |
| markmc | ttx, which bits of process do you mean? patch review, some testing efforts, ... ? | 22:12 |
| Daviey | ttx: what raised this discussion? | 22:12 |
| ttx | markmc: for example, nobody looks at the tarball contents in stable/diablo. I had to notice that things were missing. | 22:12 |
| markmc | ttx, Ubuntu and Fedora package them | 22:13 |
| ttx | markmc: and that's somethign I do as part of the dev release process | 22:13 |
| *** dprince has joined #openstack-dev | 22:13 | |
| Daviey | ttx: So what are you suggestig? | 22:13 |
| Daviey | suggesting* | 22:13 |
| ttx | I'm suggesting that distros do distributions of openstack, like they currently do | 22:13 |
| ttx | and if no distro exists for openstack on stable versions, that probably means there is no interest | 22:14 |
| Daviey | I think i missed the intro to this discussion | 22:14 |
| Daviey | context-- | 22:14 |
| ttx | and it's not openstack's job to do distro work where there is no distro | 22:14 |
| Daviey | right | 22:14 |
| markmc | ttx, whatever release process happens for dev releases, I'm happy to take on following that same process for the stable branch | 22:14 |
| ttx | it's the job of... another distro. | 22:15 |
| ttx | markmc: including time-based releases ? | 22:15 |
| markmc | ttx, sure | 22:15 |
| Daviey | ttx: well markmc is clearly motivated by Fedora interests, as we are in Ubuntu's | 22:15 |
| jog0 | @ttx, if distro's are the only one packing openstack that means hot fixes may take some time to work their way into the diablo packages? | 22:15 |
| larissa | jog0: Error: "ttx," is not a valid command. | 22:15 |
| markmc | ttx, plenty of upstream projects maintain a stable branch and do releases from it | 22:15 |
| jog0 | ttx, if distro's are the only one packing openstack that means hot fixes may take some time to work their way into the diablo packages? | 22:15 |
| markmc | ttx, and you have upstream contributors wanting to do that for OpenStack | 22:16 |
| markmc | Daviey, actually, I'm motivated by OpenStack's interests | 22:16 |
| ttx | markmc: it requires a whole new release team, if you don't want to spread current resources thin. I guess we could set that up | 22:16 |
| Daviey | markmc: purely? | 22:16 |
| markmc | Daviey, mostly, yeah | 22:16 |
| markmc | ttx, you're looking at a resource and telling him to not waste his time on it? | 22:17 |
| ttx | markmc: no, I'm making sure you understand the type of commitment you're making :) | 22:17 |
| ttx | I know from experience how much time it takes. | 22:17 |
| markmc | ttx, I think I've demonstrated I'm committed to this already | 22:17 |
| * mtaylor agrees | 22:18 | |
| markmc | ttx, maybe I should just join the release team and focus on the stable branch ? | 22:18 |
| ttx | markmc: that sounds like a very good idea. | 22:18 |
| Daviey | I think this needs documentation of what is involved first. | 22:19 |
| markmc | ttx, ok, cool stuff | 22:19 |
| ttx | another concern is the competition that this super stable branch and releases would be to the regular distros | 22:19 |
| * mtaylor nominates markmc and Kiall to become involved with stable releases | 22:19 | |
| * mtaylor does not care about that competition | 22:19 | |
| Kiall | mtaylor: I'm happy to help where I can.. | 22:20 |
| mtaylor | it is not my job to ensure that I'm not doing something that prevents a distro from being a value add | 22:20 |
| *** Mussi has joined #openstack-dev | 22:20 | |
| *** Mussi has joined #openstack-dev | 22:20 | |
| ttx | mtaylor: if it ends up being better, it would be relevant to produce packages for the maintained versions | 22:20 |
| mtaylor | it's not my job to do that for rackspace or hp either - I focus on getting stuff into the core project | 22:20 |
| ttx | mtaylor: so for example in ubuntu, would you do packages for oneiric ? | 22:21 |
| ttx | that would compete with Ubuntu's SRUs ? | 22:21 |
| markmc | ttx, there's plenty of value add that downstreams can add aside from a few months of stable branch releases from upstream | 22:21 |
| mtaylor | ttx: if we're making our packages from the same ubuntu packaging branch, I don't see what the problem is | 22:21 |
| ttx | markmc: basically I'm concerned that with PPAs on stable branches, we enter the distribution business, which we haven't done yet | 22:22 |
| ttx | markmc: (my concern about 2011.3.x releases was different) | 22:22 |
| markmc | ttx, right, I'm not arguing for OpenStack to produce any PPAs at all | 22:22 |
| ttx | markmc: (that one was just about resources) | 22:22 |
| markmc | ttx, hence me wanting to separate the discussions :) | 22:22 |
| ttx | markmc: indeed | 22:22 |
| mtaylor | I agree re: ppas | 22:23 |
| mtaylor | HOWEVER | 22:23 |
| mtaylor | I think that's a future looking thing | 22:23 |
| mtaylor | we _have_ a broken ppa for diablo | 22:23 |
| ttx | mtaylor: my fear is confusion if you do PPAs that compete with distros packages. | 22:24 |
| mtaylor | I do not fear that confusion - I think people are smart | 22:24 |
| mtaylor | I fear broken packages | 22:24 |
| ttx | mtaylor: which one should I use ? version conflicts ? where bug reports do go ? | 22:24 |
| mtaylor | how does that work right now with the release ppa? | 22:24 |
| markmc | (rather than PPAs, I'd love to see something like devstack that works with release tarballs) | 22:24 |
| ttx | mtaylor: we all agree we should remove that release PPA | 22:24 |
| ttx | mtaylor: doesn't mean we should replace it by something else | 22:25 |
| zul | ok so here is my take | 22:25 |
| mtaylor | ttx: we do? | 22:25 |
| mtaylor | ok | 22:25 |
| mtaylor | if we're going to delete the ppa, then I'm fine | 22:25 |
| ttx | release PPA was a bad idea. | 22:25 |
| mtaylor | I'm just saying if it's going to sit there, we should update it | 22:25 |
| ttx | I want to replace it by the "lasty milestone" ppa | 22:25 |
| zul | we are talking stable stuff still right? | 22:25 |
| markmc | zul, we're talking about release PPAs now | 22:25 |
| *** ayoung_ is now known as ayoung | 22:25 | |
| mtaylor | vishy: | 22:26 |
| zul | markmc: ok | 22:26 |
| ttx | mtaylor: so it's just the last milestone for test purpose, not "something you should use in prod" | 22:26 |
| mtaylor | ttx: right. I'm fine with that for our moving forward approach | 22:26 |
| mtaylor | vishy: are you ok with deleting the current release PPA? | 22:26 |
| ttx | so in summary: we should remove the release PPA, we can have 2011.3.x release provided we have specific release team members to support it | 22:27 |
| mtaylor | vishy: instead of updating it with new packages based on stable/diablo? | 22:27 |
| mtaylor | ttx: I'm fine with that | 22:27 |
| Kiall | mtaylor: what happens to people who are using/relying on that PPA? Its not like mine where people inherently dont trust it, its "offical" so people have probably relied on it. | 22:27 |
| zul | i need to do laundry but ill be back in a bit with an Ubuntu perspective | 22:27 |
| mtaylor | Kiall: that is certainly a problem | 22:27 |
| ttx | Kiall: if people use the "last milestone" ppa, they get the last milestone. | 22:28 |
| ttx | it gets updated, infrequently. | 22:28 |
| mtaylor | Kiall: however, since that ppa doesn't actually work, I doubt too many people are actually relying on it | 22:28 |
| ttx | there is no hidden promise in there | 22:28 |
| Kiall | ttx: I'm thinking about if it (the 2011.3 PPA) gets deleted that is.. rather than anything new | 22:28 |
| Kiall | mtaylor: true | 22:29 |
| *** troytoman is now known as troytoman-away | 22:29 | |
| ttx | making packages in a channel that is usable in production requires extra care -- and that's the added value that distros provide in my opinion | 22:29 |
| ttx | if we do that, we (1) do distro work and (2) probably suck at it | 22:30 |
| mtaylor | well. | 22:30 |
| *** LinuxJedi is now known as LinuxJedi|away | 22:30 | |
| mtaylor | I don't 1005 agree, of course | 22:30 |
| mtaylor | 100% | 22:30 |
| ttx | that said. | 22:30 |
| ttx | someone could do that. Someone that is not "openstack" | 22:30 |
| mtaylor | since i think the distros completely miss the use case of people on former stable distros | 22:30 |
| ttx | then it's their reponsibility if it sucks. Not ours | 22:30 |
| mtaylor | and I think that is why tons of projects provide packages | 22:30 |
| mtaylor | it's not their responsiblity if someone's experience of our code sucks | 22:31 |
| mtaylor | it's ours | 22:31 |
| mtaylor | it's one of the reaons that mysql punked postgresql for so long | 22:31 |
| mtaylor | mysql was installable | 22:31 |
| mtaylor | postgres was not | 22:31 |
| mtaylor | so people used mysql | 22:31 |
| mtaylor | this stuff matters | 22:31 |
| ttx | mtaylor: I think we would do a bad job at it -- or lose a lot of resources around it | 22:31 |
| mtaylor | ttx: I think if we can't figure out pacakging, then we should get out of the software business personally | 22:32 |
| ttx | it also creates conflicts with distros | 22:32 |
| mtaylor | but I agree that without the will to do it | 22:32 |
| mtaylor | I do not care about conflicts with distros | 22:32 |
| markmc | mtaylor, also, any upstream releasing packages really has to get into the game of doing it for multiple platforms/distros | 22:32 |
| ttx | mtaylor: doing a production channel is a bit more than just "packaging" | 22:32 |
| mtaylor | markmc: yes. they do | 22:32 |
| mtaylor | I've been arguing for that for months, actually :) | 22:32 |
| jeblair | mtaylor: essex will be installable in precise, and that will be supported long term. doesn't that address your issue? | 22:33 |
| ttx | mtaylor: see what the SRU team does in Ubuntu | 22:33 |
| mtaylor | jeblair: not really | 22:33 |
| mtaylor | jeblair: it's a step | 22:33 |
| mtaylor | but it's only going to be relevant for a while | 22:33 |
| mtaylor | because then what happens with essex+3 comes out | 22:33 |
| mtaylor | is ubuntu going to be following latest openstack releases on precise? | 22:34 |
| ttx | mtaylor: so we'll do packaging and pray that it is usable in production, and not really test version-to-version upgrades or whatnot | 22:34 |
| mtaylor | ttx: no, we need to test version to version upgrades | 22:34 |
| markmc | mtaylor, could we release to pypi and have people consume from there, perhaps with an easy installer provided by OpenStack? | 22:34 |
| jeblair | i would expect that precise+3 would package essex+3. | 22:34 |
| mtaylor | markmc: no. pypi is DRASTICALLY unusable for a production deployment | 22:34 |
| mtaylor | jeblair: right. and then how are the users with datacenters running precise who want to run essex+3 served | 22:35 |
| ttx | mtaylor: that takes resources -- then I say, we lose focus on the development release. I suspect you're arguing it's as important if not more ? | 22:35 |
| jeblair | mtaylor: they run datacenters. they are smart enough to backport packages. | 22:35 |
| jeblair | mtaylor: http://iuscommunity.org/ | 22:35 |
| mtaylor | I mean - I'm obviously outnumbered, and that's fine. I'm just saying that the distros do NOT solve all of the use cases | 22:36 |
| mtaylor | so when we say "get it from the distros" - that's not what we actually mean | 22:36 |
| notmyname | I can't make the arguments as well or type as fast, but I agree with mtaylor | 22:36 |
| mtaylor | what we mean is "we expect most of our users to backport and build packages themselves" | 22:36 |
| Kiall | notmyname: I know that feeling ;) | 22:37 |
| mtaylor | I think we should just say that, if that's what we're saying | 22:37 |
| mtaylor | I disagree with it | 22:37 |
| jeblair | mtaylor: i agree with that. but i also believe that if we want to solve that problem, we need to put the kind of resources into the problem that distros do. | 22:37 |
| jeblair | mtaylor: isn't that what we said at the summit? | 22:37 |
| mtaylor | yes. I think what I'm saying is that we should solve the problem and that we should put resources in to it | 22:37 |
| mtaylor | at the moment, we are punting on it | 22:37 |
| jeblair | mtaylor: isn't that what "packages are not a first class deliverable" means? | 22:37 |
| mtaylor | but I think it's something that should be addressed | 22:37 |
| mtaylor | jeblair: yes. it is. | 22:37 |
| Daviey | markmc: When did we decide to do point releases? I thought */stable was always in a release state? | 22:37 |
| jeblair | mtaylor: okay. so are you trying to convince people in this room to spend time working on it? | 22:38 |
| zul | umm...hi...speaking as a distro | 22:38 |
| mtaylor | jeblair: nope. i'm saying that people are already working on it, so that might point to people wanting it to exist | 22:38 |
| *** bcwaldon has quit IRC | 22:38 | |
| zul | we are in the process of doing a stable/diablo SRU for oneiric we are getting through the beurorcary | 22:38 |
| rmk | stable/diablo should be the foundation of point releases, imo | 22:38 |
| *** apevec has quit IRC | 22:38 | |
| jeblair | mtaylor: who's working on it? | 22:38 |
| mtaylor | jeblair: Kiall, for one | 22:38 |
| markmc | Daviey, certainly a goal not to introduce regressions, but you do need to actually bless checkpoints IMHO | 22:39 |
| zul | mtaylor Kiall ttx: ^^^ | 22:39 |
| Daviey | markmc: Right, so we are putting significant effort into per stable testing.. so we can treat each commit as release ready. | 22:39 |
| markmc | Daviey, we'll make mistakes occasionally, a 2011.3.x release process will help increase the likelihood of catching them | 22:39 |
| mtaylor | jeblair: Kiall is maintaining a PPA. If there isn't an official one, his is going to become the unofficial offical one soon enough - and if it doesn't work right, people are still going to assume the fault is with us | 22:39 |
| zul | if you do a lucid backport then thats fine but you are going to be inheriting a hell of alot debt if you decide to do that | 22:39 |
| notmyname | mtaylor: as the rackspace one for swift is "unofficially official" | 22:40 |
| jeblair | Kiall: what are you supporting in your ppa? | 22:40 |
| mtaylor | exactly | 22:40 |
| markmc | Daviey, "we" == Ubuntu? that's great, gives upstream greater confidence about doing releases from the branch | 22:40 |
| Daviey | markmc: I understood keystone was special cased because the /release/ wasn't solid. | 22:40 |
| rmk | mtaylor: That's exactly what's happening btw. | 22:40 |
| Kiall | jeblair: just oneiric, i figure the precise LTS is close enough to ignore the effort of lucid. | 22:40 |
| mtaylor | rmk: I know. that's why I'm arguing that all of our current theory is BS, because in practice the situation is quite different | 22:41 |
| *** lts has quit IRC | 22:41 | |
| markmc | Daviey, right, I see keystone as being outside the stable branch team's responsibility for diablo, since it was incubating | 22:41 |
| rmk | mtaylor: Agreed. | 22:41 |
| Daviey | markmc: the 'we's are getting a little messy. | 22:41 |
| markmc | Daviey, it certainly didn't follow the wiki.openstack.org/StableBranch process | 22:41 |
| rmk | mtaylor: I thought you were opposed to maintaining stable packages but it sounds like it's the opposite. | 22:41 |
| mtaylor | rmk: I'm opposed to maintaining stable packages without resources | 22:41 |
| rmk | Fair enough | 22:41 |
| ttx | mtaylor: I've trouble following you. Recently you wanted to remove packages from release deliverables and now you're into maintaining a production packages channel | 22:42 |
| mtaylor | rmk: or in a completely ad-hoc manner | 22:42 |
| ttx | mtaylor: the arguments you used then could be used against you now | 22:42 |
| mtaylor | ttx: totally | 22:42 |
| mtaylor | ttx: what I'm saying is that defacto official packages are springing up | 22:42 |
| notmyname | ttx: my understanding is that mtaylor didn't want us to have packages without an official commitment to support them | 22:42 |
| jeblair | notmyname: regardless of whatever mtaylor currently believes, i agree with the sentiment you just expressed. :) | 22:43 |
| notmyname | lol | 22:43 |
| rmk | I've said this before and I'll say it again, Openstack already gets a lot of fuss around "how difficult it is to install". The current packaging situation is not helping the matter at all. | 22:43 |
| mtaylor | ttx: and that we might need to accept that as a fact of life | 22:43 |
| mtaylor | ttx: and deal with it | 22:43 |
| notmyname | jeblair: I agree too, but I think openstack should provide supported packages | 22:43 |
| rmk | Having a stable set of packages seems critical to that. | 22:43 |
| mtaylor | ttx: rather than continuing down the road of "we're not going to deal with this" | 22:43 |
| ttx | notmyname: for all Linux distributions ? | 22:44 |
| * mtaylor agrees with notmyname | 22:44 | |
| ttx | if not, for which ? | 22:44 |
| rmk | ttx: At the very least for one distro. | 22:44 |
| notmyname | ttx: probably for ubuntu + RHEL (+ maybe debian) | 22:44 |
| zul | ttx: ok so speaking as someone who is working on Ubuntu we *are* doing a stable release tree for oneiric, we are working on a testing infrastructure to test out stable/diablo on oneiric, its just ready yet | 22:44 |
| mtaylor | and thinks that ubuntu, debian, fedora and centos should be supported at the very least | 22:44 |
| rmk | But yeah I'd say Ubuntu + RHEL. | 22:44 |
| zul | mtaylor: good luck with that especially for backports | 22:44 |
| mtaylor | zul: thanks! | 22:45 |
| mtaylor | zul: I'm not saying it's going to be fun or easy - just that it's work that's going to happen somewhere, so we might as well play a hand in coordinating it | 22:45 |
| mtaylor | zul: otherwise, every company that's deploying openstack is going to duplicate all of that backport work | 22:45 |
| ttx | mtaylor: saying we should do it won't make it happen. With our current resources we'll do a very bad job at it. I'd better have nothing than an half-assed job branded "openstack" | 22:46 |
| markmc | mtaylor, wait, *every* company implies no companies consume OpenStack from distros? | 22:46 |
| zul | sure but you have to worry about things like libvirt and its just not good enough to say libvirt builds on lucid it has to actually run and who is going to maintain the security updates | 22:46 |
| mtaylor | zul: sure. but if people are running lucid | 22:47 |
| ttx | mtaylor: my point is that it's a SEPARATE effort from openstack. It's a distribution. A well-needed one | 22:47 |
| mtaylor | zul: then they are going to be forced to do that work | 22:47 |
| ttx | it will take SEPARATE people to do it and succeed | 22:47 |
| mtaylor | zul: and I'd rather give them a location where they can collaborate on it | 22:47 |
| ttx | members of openstack can certainly be in the two efforts | 22:47 |
| ttx | mtaylor: but in the end, distro work is different from upstream work | 22:47 |
| zul | mtaylor: if the openstack project does that they they will need a lot of resources and that will take away from essex | 22:47 |
| markmc | ttx, well, if such an effort was successful, it could be part of openstack | 22:48 |
| notmyname | mtaylor: coordinating the work that rackspace + nebula + piston + et al are all doing? | 22:48 |
| mtaylor | no. distro work is different from upstream work because of the way that distros are organized | 22:48 |
| mtaylor | notmyname: ++ | 22:48 |
| markmc | ttx, I think the point would be that openstack wouldn't want a half-assed attempt at it | 22:48 |
| markmc | ttx, maybe such an effort should go through incubation too, in theory? | 22:48 |
| mtaylor | I have no vested interest in maintaining or supporting the model of the current distros when it doesn't solve the problems of my users | 22:48 |
| *** dprince has quit IRC | 22:49 | |
| mtaylor | even when I'm friends with the people working there | 22:49 |
| ttx | mtaylor: you're not hearing me. You're doing a new distro. Not just a package. | 22:49 |
| jeblair | mtaylor: your typical example for this need is lucid. is anyone maintaining a lucid ppa? | 22:49 |
| mtaylor | ttx: openstack is a distro already | 22:49 |
| ttx | mtaylor: so what you ask for is that openstack provides its own distro. | 22:49 |
| jeblair | mtaylor: who are your users? | 22:49 |
| ttx | or rather, a set of distros | 22:49 |
| *** bsza has quit IRC | 22:50 | |
| notmyname | jeblair: rackspace's swift deployment is all on lucid | 22:50 |
| mtaylor | jeblair: we've been carrying lucid in the openstack ppas already | 22:50 |
| mtaylor | jeblair: and lucid is in the openstack release ppa right now | 22:50 |
| rmk | A release PPA no one should use honestly. | 22:50 |
| jeblair | mtaylor: that's the unsupported thing we want to get rid of, right? | 22:51 |
| ttx | and we all agree that it should vanish, let's not rehash that | 22:51 |
| Daviey | openstack does touch too much of the ecosphere to consider it just a project, like any other. | 22:51 |
| rmk | Either vanish or be maintained. | 22:51 |
| mtaylor | what rmk said. | 22:51 |
| mtaylor | we want to deal with the problem that it is unsupported. one of the ways to do that is to delete it. one of the ways to do that is to support it. | 22:51 |
| ttx | mtaylor: so what you're suggesting is that openstack expands its scope to enter the distribution business | 22:52 |
| mtaylor | ttx: what I am arguing is that we have been in the distribution business since day one | 22:52 |
| ttx | mtaylor: that's a discussion for the PPB -- and I don't see why it's not relevant for a ML discussion as well | 22:52 |
| mtaylor | ttx: and that we should acknowledge that fact formally | 22:52 |
| *** dolphm has quit IRC | 22:52 | |
| ttx | mtaylor: our packages were never meant to be used in production, though. We relied on our ecosystem to do that | 22:53 |
| Daviey | I think first, we need to divide up what is being discussed | 22:53 |
| Daviey | there is at least 3 different topics being discussed concurrently | 22:53 |
| ttx | mtaylor: expanding the scope of the project is not a neutral decision | 22:53 |
| mtaylor | ttx: it's not expansion. just because we say that those weren't meant to be used in production doesn't make it even close to being accurate | 22:53 |
| ttx | mtaylor: you can't take it all by yourself in a IRC discussion, so I still think this warrants a ML discussion at least, probably a PPB decision | 22:53 |
| mtaylor | sure thing | 22:54 |
| Daviey | +1 | 22:54 |
| mtaylor | although I predict that nothing useful will come from the ML discussion of it | 22:54 |
| ttx | mtaylor: you're shutting down a few business models around openstack, at the expense of diverting project resources. That's not neutral. | 22:54 |
| mtaylor | ttx: I do not need to support anti-features | 22:54 |
| Daviey | anti-features? | 22:54 |
| ttx | mtaylor: People at Piston or Canonical or Nebula may disagree with you | 22:55 |
| mtaylor | if the business models are based on project deficiences, I do not need to avoid fixing them | 22:55 |
| mtaylor | I think canonical's business model is just fine | 22:55 |
| mtaylor | as is nebula's | 22:55 |
| mtaylor | and I'm pretty sure piston's is different too | 22:55 |
| ttx | no, business models are based on what the project deliberately chose not to focus on. | 22:55 |
| openstackgerrit | Don Dugger proposed a change to openstack-dev/devstack: Enhanced identification of HOST_IP and eth device https://review.openstack.org/1953 | 22:56 |
| Daviey | Hang on, i'm not sure different companies business models is a relevant discussion in the develoment channel. | 22:56 |
| ttx | mtaylor: this has been discussed in the past. | 22:56 |
| jeblair | https://lists.launchpad.net/openstack/msg04789.html | 22:56 |
| jeblair | that's mtaylor's excellent summary of the discussions at ods | 22:56 |
| jeblair | for those who would like a refresher | 22:56 |
| Daviey | This is not something any of us present are qualified to be a pundit on | 22:56 |
| Daviey | (myself included) | 22:56 |
| mtaylor | Daviey: ++ | 22:57 |
| ttx | mtaylor: as in, around the start of Bexar, when we set the limits | 22:57 |
| jeblair | https://lists.launchpad.net/openstack/msg04835.html is pretty much the conversation we're having now. | 22:57 |
| ttx | those limits were set to allow an ecosystem to grow around openstack, and to focus the project on what it does best. | 22:57 |
| ttx | i.e. coding. Not packaging or maintaining a production distro. | 22:58 |
| mtaylor | ttx: ok. so what do you do about unofficial official repos. just ignore them? | 22:58 |
| ttx | mtaylor: that's great. I wish one of them wil be successful, a commpany formed around it and the one at the top of it being rich selling openstack expertise. | 22:59 |
| markmc | IMHO, *iff* a thriving community effort existed around distributing OpenStack on multiple platforms, the project couldn't shun it forever | 22:59 |
| mtaylor | why does all of this have to be about MONEY???? | 22:59 |
| Daviey | why would you do anything else? | 22:59 |
| mtaylor | is no one else here hacking on an open source project? | 22:59 |
| mtaylor | does no one else just want the project to be successful? | 22:59 |
| mtaylor | why does packaging have to be tied to a rick person making money off of a business model | 22:59 |
| Daviey | mtaylor: that is kinda emotive, no? | 23:00 |
| ttx | mtaylor: so that we can focus on what we do best | 23:00 |
| mtaylor | can we? | 23:00 |
| ttx | mtaylor: they don't have to make money out of it | 23:00 |
| ttx | mtaylor: they could do "SOS Linux" as a free-for-all community | 23:01 |
| ttx | "Stable Open Stack" | 23:01 |
| openstackgerrit | Brian Waldon proposed a change to openstack/nova: Move users and accounts from OSAPI into Admin API https://review.openstack.org/1954 | 23:01 |
| ttx | mtaylor: mind you, I'm not saying we shouldn't do it -- I'm just saying it's seriously stretching the limits we set up early on. | 23:02 |
| *** ayoung has quit IRC | 23:02 | |
| ttx | mtaylor: limits, like I said, that helped grow an ecosystem of projects and distros around openstack, while keeping us focused on coding | 23:02 |
| Daviey | mtaylor: It's really not clear to me what you are suggesting. | 23:03 |
| ttx | mtaylor: so that's not a neutral decision. | 23:03 |
| ttx | and on that, I'll go to bed | 23:03 |
| mtaylor | Daviey: really? I'm quite clearly suggesting that we should be producing packages for the releases of distros that aren't be served by the distros themselves | 23:04 |
| mtaylor | Daviey: I'm saying that a project that can't be installed is a non-starter for loads of folks | 23:04 |
| ttx | markmc: I still need to be convinced of the value of point releases, btw. I see how we can do them now, but I still don't see why. | 23:04 |
| Kiall | "so what do you do about unofficial official repos. just ignore them?" The reason I re-packaged everything was because the Ubuntu packages were (are?) broken. zul mentioned they we're getting fixed, but are held up in "bureaucracy".. Frankly, I'd love if Ubuntu was able to provide solid, up to date packages. I know I'm not the only one who wants that. Anyway - Just my 2c on why I repackaged OS... | 23:04 |
| Daviey | mtaylor: Hmm, i thought you pushed hard for that to stop - against what we were asking for? :) | 23:04 |
| mtaylor | Daviey: I pushed hard to not produce packages if they weren't going to be supported | 23:05 |
| markmc | ttx, I'll let you go to bed :) another time | 23:05 |
| rmk | ttx: When users are reporting bugs or issues in openstack "stable", how will you know specifically which fixes/patches they are using? | 23:05 |
| Daviey | Kiall: We welcome more collaboration fwiw. | 23:05 |
| ttx | Kiall: bureaucracy is there for a reason, too | 23:05 |
| Daviey | mtaylor: are you signing up to support them? | 23:05 |
| *** kbringard has quit IRC | 23:05 | |
| rmk | "I'm using stable as of 10/22" isn't really a clear method of communicating that. | 23:05 |
| *** markvoelker has quit IRC | 23:06 | |
| ttx | Kiall: that's the part where doing a production-ready distro is a bit more work than just packaging. | 23:06 |
| Daviey | markmc: Wait, go to bed now - and you'll find the project has been renamed by the morning. | 23:06 |
| Daviey | probably re-written in ruby aswell :) | 23:06 |
| Kiall | ttx: I know, but the delays introduced bu the bureaucracy, while there were broken packages and users trying to install was a failure. Even if that bureaucracy usually means better packages. | 23:06 |
| *** dprince has joined #openstack-dev | 23:07 | |
| ttx | Kiall: so that's what we are after ? Quicker fixes at the expense of production-readiness ? | 23:07 |
| Kiall | ttx: not at all. | 23:07 |
| novas0x2a|laptop | for what it's worth, piston has a hack solution for the "what version are you on" problem: https://github.com/piston/nova/commit/ccb9d91ca0be48a98dd99daee4836c551c09c837#L2L42 (we build PISTON_VERSION as 9999-<BUILDID>-<GITREVISION> | 23:08 |
| novas0x2a|laptop | i'd be glad to figure out how to refine it for upstream if people want that | 23:08 |
| Kiall | the packages were not production ready to begin with! I believe that was due to the 2011.3 release date being so close to oneiric. | 23:08 |
| novas0x2a|laptop | ... closing my dangling ) | 23:08 |
| *** markmc has quit IRC | 23:08 | |
| mtaylor | novas0x2a|laptop: I am infavor of including version meta-information | 23:09 |
| rmk | No one in their right mind is going to do anything with the official 2011.3 packages. | 23:10 |
| * ttx really goes to bed before saying anything silly. | 23:10 | |
| Kiall | Anyway - Its 11PM and I'm still in the office... Its time to head home! | 23:10 |
| mtaylor | Kiall, ttx: get some sleep! | 23:10 |
| novas0x2a|laptop | mostly the problem is that nova and glance do versions one way, keystone does another, and swift does another. i just sort of bludgeoned stuff together to get it to work the way we wanted. heh. | 23:10 |
| novas0x2a|laptop | i hate how setuptools punts on the version problem :( | 23:10 |
| * Daviey ponders where Kiall is based | 23:10 | |
| mtaylor | novas0x2a|laptop: omg yes | 23:10 |
| Kiall | Dublin, Irealnd.. | 23:10 |
| mtaylor | novas0x2a|laptop: I hate how setuptools punts on several things | 23:11 |
| Daviey | Ahh | 23:11 |
| Kiall | Ireland* (Its late!) | 23:11 |
| Daviey | right, 23:00 here aswell | 23:11 |
| Daviey | nn | 23:11 |
| novas0x2a|laptop | nity | 23:11 |
| Kiall | cyas! | 23:11 |
| ttx | and 00:11 here. | 23:11 |
| mtaylor | novas0x2a|laptop: I've been working on a version module for python based on some of the things we do in openstack jenkins jobs, and also in work that I did for drizzle a couple of years ago - end goal for that would be direct setuptools integration | 23:11 |
| mtaylor | novas0x2a|laptop: maybe we should chat about it - I'd love feedback/help | 23:12 |
| novas0x2a|laptop | that'd be great :D | 23:12 |
| novas0x2a|laptop | sure | 23:12 |
| *** dprince has quit IRC | 23:14 | |
| vishy | wow | 23:14 |
| vishy | I blew up openstack-dev with my wee little suggestion | 23:15 |
| vishy | all i wanted was a ppa that builds off of stable/diablo that I can tell people to use | 23:15 |
| novas0x2a|laptop | 24/7 powder keg XD | 23:15 |
| mtaylor | vishy: you should learn to not say things :) | 23:16 |
| vishy | :| | 23:16 |
| rmk | vishy: Which is completely reasonable and wanted by many but is getting quite a bit of push back. | 23:17 |
| jog0 | the gist of the conversation sounded like no one else liked that idea. It was either no release ppa or highly tested ppa. | 23:17 |
| jeblair | vishy: in the spirit of data collection: who do you want to use that ppa? | 23:17 |
| rmk | As it stands, you've got a completely useless release PPA with packages no one should use. | 23:17 |
| *** Mandell_ has joined #openstack-dev | 23:17 | |
| vladimir3p | folks, it seems like we are ending up completely without usefull ppas | 23:17 |
| vladimir3p | for example, for us we can't use essex, because they are unstable | 23:18 |
| rmk | Honestly, I'm surprised the project has gone this long without having a point release methodology | 23:18 |
| rmk | In essense, you can rest assured that every single deployment of Openstack is different. | 23:18 |
| *** jeffjapan has joined #openstack-dev | 23:18 | |
| vladimir3p | the official diablo release is broken, as you know. So, now every company will need to either use developer's envs or .. build packages by their own | 23:18 |
| *** kimscheibel_ has joined #openstack-dev | 23:19 | |
| *** lloydde has quit IRC | 23:19 | |
| rmk | I don't even know how a distro deals with this when there isn't a concise stable release strategy (beyond just having a stable branch). | 23:19 |
| *** bencherian has quit IRC | 23:20 | |
| *** pixelbeat has joined #openstack-dev | 23:21 | |
| *** bencherian has joined #openstack-dev | 23:21 | |
| jog0 | what is so broken about the 2011.3 PPA? | 23:24 |
| *** rnirmal has quit IRC | 23:24 | |
| *** lloydde has joined #openstack-dev | 23:28 | |
| jog0 | vladimir3p, ^^ | 23:29 |
| vishy | jog0: https://github.com/openstack/nova/compare/2011.3...stable/diablo | 23:30 |
| jog0 | so its just very out of date with known bugs? | 23:31 |
| vishy | jog0: | 23:33 |
| vishy | yes there are a few that are real production issues | 23:33 |
| vishy | like 1/10 failures when running concurrent instances | 23:34 |
| vishy | etc. | 23:34 |
| jog0 | and this is why you want to just have a PPA based on stable | 23:34 |
| jog0 | vishy, thanks for explaining | 23:35 |
| vishy | jog0: yes | 23:35 |
| vishy | jog0: people are worried about having to support it | 23:35 |
| vishy | but i think just having code drops is hurting the community a bit | 23:35 |
| jog0 | wouldn't supporting it just be doing the regular back ports of bug fixes? | 23:36 |
| vishy | i don't want packages for every distro | 23:36 |
| vishy | jog0: we have a branch for that | 23:36 |
| vishy | it is maintaining the dependencies and packages that is hard | 23:36 |
| vishy | well apparently hard | 23:36 |
| vishy | we had it working just fine for a long time so I don't think it is as hard as people are suggesting. | 23:37 |
| vishy | although that was mostly because people like mtaylor, soren, Daviey were on top of it. | 23:37 |
| jog0 | yeah, I am not sure why minor bug fixes would change the dependancies so much | 23:37 |
| soren | jog0: You'd be surprised. | 23:38 |
| jog0 | soren: do you have an example of the challenges you had in the past when you were doing it? | 23:39 |
| soren | jog0: I don't know what the challenges were, but people have done it numerous times. | 23:39 |
| soren | jog0: My brain has removed the memories to maintain my sanity. | 23:40 |
| *** bencherian has quit IRC | 23:40 | |
| rmk | Is there any harm in running 2+ schedulers in a given availability zone? Wondering if there's a nomination process via rabbit to determine who is actually handling scheduling. | 23:41 |
| *** bencherian has joined #openstack-dev | 23:42 | |
| vishy | rmk: the nomination process is the first worker to grab the message handles it | 23:44 |
| rmk | Are there any senarios where having multiple schedulers can break things? | 23:45 |
| *** code_franco has quit IRC | 23:46 | |
| vishy | rmk: not that i am aware of | 23:46 |
| rmk | great, thanks | 23:46 |
| *** pixelbeat has quit IRC | 23:47 | |
| *** GheRivero has quit IRC | 23:49 | |
| jog0 | soren, vishy: While I understand the hesitation to put the resources behind running a good stable nova PPA, when the foundation starts up will they cover this role? | 23:50 |
| vishy | jog0: maybe | 23:51 |
| openstackgerrit | Dean Troyer proposed a change to openstack-dev/devstack: Set sane defaults, get config info from localrc https://review.openstack.org/1955 | 23:51 |
| vishy | jog0: some people feel that we should stay out of packaging completely | 23:51 |
| *** adjohn has joined #openstack-dev | 23:53 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!