| *** adiantum has joined #openstack-dev | 00:00 | |
| *** lorin1 has quit IRC | 00:01 | |
| *** nati has quit IRC | 00:04 | |
| *** adiantum has quit IRC | 00:12 | |
| *** dabo has quit IRC | 00:16 | |
| *** novas0x2a|laptop has quit IRC | 00:16 | |
| *** jdurgin has quit IRC | 00:16 | |
| *** bengrue has quit IRC | 00:16 | |
| *** rnorwood has joined #openstack-dev | 00:17 | |
| *** bengrue has joined #openstack-dev | 00:17 | |
| *** adiantum has joined #openstack-dev | 00:18 | |
| *** lorin1 has joined #openstack-dev | 00:19 | |
| *** nati has joined #openstack-dev | 00:22 | |
| *** lorin1 has quit IRC | 00:23 | |
| *** adiantum has quit IRC | 00:24 | |
| *** nati has quit IRC | 00:25 | |
| *** nati has joined #openstack-dev | 00:25 | |
| *** deshantm_away is now known as deshantm | 00:25 | |
| *** bengrue has quit IRC | 00:26 | |
| *** nati_ has joined #openstack-dev | 00:28 | |
| *** nati has quit IRC | 00:29 | |
| *** adiantum has joined #openstack-dev | 00:29 | |
| *** bengrue has joined #openstack-dev | 00:31 | |
| *** nati_ has quit IRC | 00:35 | |
| *** dabo has joined #openstack-dev | 00:36 | |
| *** adiantum has quit IRC | 00:39 | |
| *** nati has joined #openstack-dev | 00:40 | |
| *** JordanRinke is now known as JordanRinke-AFK | 00:42 | |
| *** troytoman is now known as troytoman-away | 00:44 | |
| *** adiantum has joined #openstack-dev | 00:45 | |
| chmouel | is userdata exposed to the OS api ? from a quick grep on api/openstack/servers.py it does not seems so (and does not show on the req) | 00:47 |
|---|---|---|
| *** lorin1 has joined #openstack-dev | 00:50 | |
| *** adiantum has quit IRC | 01:02 | |
| *** adiantum has joined #openstack-dev | 01:03 | |
| *** lorin1 has quit IRC | 01:07 | |
| *** lorin1 has joined #openstack-dev | 01:08 | |
| *** adiantum has quit IRC | 01:10 | |
| *** rnorwood has quit IRC | 01:13 | |
| *** rnorwood has joined #openstack-dev | 01:15 | |
| *** adiantum has joined #openstack-dev | 01:17 | |
| *** thickskin has left #openstack-dev | 01:22 | |
| *** adiantum has quit IRC | 01:24 | |
| *** mattray has joined #openstack-dev | 01:24 | |
| vishy | chmouel: sure is | 01:24 |
| vishy | well depends on what you mean by exposed? | 01:25 |
| chmouel | so i store fine user_data stuff when i do the POST | 01:25 |
| chmouel | and it's in DB | 01:25 |
| chmouel | but when I try to retrieve later it does not show anywhere | 01:25 |
| chmouel | maybe this is not the right use of user-data | 01:25 |
| vladimir3p | vishy: ping | 01:25 |
| vishy | hoyo | 01:26 |
| vishy | chmouel: it is available to the instance | 01:26 |
| vishy | from the metadata server | 01:26 |
| vishy | if you want data for the api | 01:26 |
| vishy | you want to use instance meta property | 01:26 |
| vladimir3p | vishy: hi, any tip on debugging amqp? I have a very interesting situation right now in one of environments where funout msg doesn't reach one of nodes | 01:26 |
| vladimir3p | *fanout | 01:26 |
| vishy | vladimir3p: hmm, not sure | 01:27 |
| chmouel | okay so what user_data is supposed to do compared to metadata? maybe i have the wrong understanding of it | 01:27 |
| vladimir3p | vishy: np, thanks | 01:28 |
| vishy | user_data is b64 encoded text that the instance can retrieve via 169.254.169.254 | 01:28 |
| vishy | it is used by the cloud-init scripts in ubuntu for example | 01:28 |
| chmouel | ie: file injection? | 01:29 |
| vishy | metadata is arbitrary key value pairs associated with the instance | 01:29 |
| vishy | well the files aren't injected | 01:29 |
| vishy | it can be retrived by hitting that special ip | 01:29 |
| *** adiantum has joined #openstack-dev | 01:29 | |
| chmouel | yep so they can be accecible to VM okay gotcha | 01:30 |
| *** thickskin has joined #openstack-dev | 01:30 | |
| *** rods has quit IRC | 01:30 | |
| vishy | chmouel: right. To get metadata, you go through the api, so you need credentials | 01:31 |
| chmouel | vishy: okay great thanks, i was trying to understanding how it works so I can finish jesse's initial userdata support in python-novaclient http://goo.gl/a2cOA | 01:32 |
| chmouel | don't think i have much to add to that except unittests | 01:33 |
| chmouel | and maybe s userdata user_data like named on nova code | 01:33 |
| *** adiantum has quit IRC | 01:35 | |
| *** adiantum has joined #openstack-dev | 01:41 | |
| *** nati has quit IRC | 01:42 | |
| *** mfer has joined #openstack-dev | 01:43 | |
| *** bengrue has quit IRC | 01:45 | |
| *** adiantum has quit IRC | 01:48 | |
| *** lorin1 has quit IRC | 01:49 | |
| *** vladimir3p has quit IRC | 01:49 | |
| *** adiantum has joined #openstack-dev | 01:53 | |
| *** lorin1 has joined #openstack-dev | 01:54 | |
| *** mfer has quit IRC | 01:57 | |
| *** adiantum has quit IRC | 02:00 | |
| *** adiantum has joined #openstack-dev | 02:13 | |
| openstackgerrit | Justin Shepherd proposed a change to openstack/glance: Bug fix lp:726864 https://review.openstack.org/367 | 02:19 |
| openstackgerrit | A change was merged to openstack/glance: Bug fix lp:726864 https://review.openstack.org/367 | 02:22 |
| *** mwhooker has quit IRC | 02:23 | |
| *** dragondm has quit IRC | 02:23 | |
| *** mdomsch has joined #openstack-dev | 02:29 | |
| *** mattray has quit IRC | 02:42 | |
| *** novas0x2a|laptop has joined #openstack-dev | 02:55 | |
| *** ewindisch has quit IRC | 03:04 | |
| *** tsuzuki_ has joined #openstack-dev | 03:09 | |
| vishy | comstud: ping | 03:15 |
| *** lorin1 has quit IRC | 03:17 | |
| *** novas0x2a|laptop has quit IRC | 03:26 | |
| *** martine_ has quit IRC | 03:32 | |
| *** novas0x2a|laptop has joined #openstack-dev | 03:34 | |
| comstud | vishy: pong | 03:35 |
| vishy | comstud: made a couple comments on rpc | 03:35 |
| vishy | doesn't work with kombu 1.0.4 | 03:35 |
| *** rnorwood has quit IRC | 03:36 | |
| comstud | vishy: bummer.. i forgot. i wanted to test various versions.. | 03:36 |
| comstud | i think the minimal one i was using was 1.1.2 or something | 03:36 |
| comstud | works with 1.1.6 for sure | 03:37 |
| comstud | that's on one of my dev boxes I was doing functional testing with | 03:37 |
| vishy | ok | 03:37 |
| vishy | so maybe 1.1+ | 03:37 |
| comstud | yeah, maybe.. what was broken? | 03:38 |
| vishy | we still need packages for the ppa | 03:38 |
| comstud | *pulls up comments* | 03:38 |
| vishy | looks like there is no Connection class | 03:38 |
| vishy | :) | 03:38 |
| comstud | hm | 03:38 |
| comstud | ah, yeah, i'm using kombu.connection.Connection | 03:39 |
| comstud | maybe I can fix it for 1.0.4 | 03:39 |
| comstud | our ppa pkg is 1.0.4? | 03:39 |
| vishy | well the one for glance is/was | 03:40 |
| comstud | ok | 03:40 |
| comstud | i'll take a look at 1.0.4 and see what the connection class there is | 03:40 |
| comstud | or | 03:40 |
| comstud | i'll look at glance | 03:40 |
| comstud | BrokerConnection, eh | 03:41 |
| comstud | do you still have a 1.0.4 install? | 03:41 |
| comstud | looks like 1.1.2 has a Connection = BrokerConnection in connection.py | 03:42 |
| comstud | so if I change to use kombu.connection.BrokerConnection this will still work in 1.1.2... wonder if that's all that will make 1.0.4 work | 03:42 |
| vishy | i can check it | 03:43 |
| vishy | all i have to do is change Connection to BrokerConnection? | 03:43 |
| comstud | yeah | 03:44 |
| comstud | line 325 | 03:44 |
| comstud | i'll revert to the ppa version and see what gives here too | 03:45 |
| openstackgerrit | Brian Waldon proposed a change to openstack/glance: correcting recent changes https://review.openstack.org/368 | 03:46 |
| openstackgerrit | Brian Waldon proposed a change to openstack/glance: correcting recent changes https://review.openstack.org/368 | 03:47 |
| comstud | kombu tests pass | 03:47 |
| comstud | with the change | 03:47 |
| comstud | >>> kombu.__version__ | 03:47 |
| comstud | '1.0.4' | 03:47 |
| openstackgerrit | Brian Waldon proposed a change to openstack/glance: Fix cached-images API endpoint https://review.openstack.org/368 | 03:49 |
| *** rnorwood has joined #openstack-dev | 03:49 | |
| comstud | testing functionally | 03:49 |
| comstud | seems to work | 03:50 |
| vishy | comstud: works fine with BrokerConnection here too | 03:50 |
| vishy | does it work with the new code? | 03:50 |
| vishy | as well | 03:50 |
| vishy | as in 1.2.1/BrokerConnection | 03:50 |
| comstud | what do you mean by 'new code' ? | 03:50 |
| comstud | oh | 03:50 |
| comstud | it should.. as Connection = BrokerConnection in new code | 03:51 |
| comstud | i'll test again just in case | 03:51 |
| *** novas0x2a|lapto1 has joined #openstack-dev | 03:51 | |
| vishy | i'm checking too | 03:52 |
| comstud | tests pass | 03:52 |
| comstud | for kombu | 03:52 |
| comstud | seems to work functionally as well | 03:53 |
| comstud | this is with 1.2.1 | 03:53 |
| *** novas0x2a|laptop has quit IRC | 03:53 | |
| comstud | pushing up fix | 03:54 |
| vishy | yup | 03:54 |
| vishy | agreed | 03:54 |
| vishy | as far as I can tell its sexy time | 03:54 |
| comstud | i like sexy time | 03:55 |
| *** rnorwood has quit IRC | 03:56 | |
| vishy | maybe we can switch the default | 03:57 |
| vishy | we can always revert if there are any big issues | 03:57 |
| comstud | up to you | 03:57 |
| comstud | switching the default forces people to use it | 03:57 |
| comstud | and we'll shake things out that way if there's anything | 03:58 |
| comstud | also makes these 2 other bugs go away.. which I didn't bother trying to mess with carrot code to fix | 03:59 |
| comstud | I feel fixing those bugs in carrot makes it almost as risky as just using kombu. Proper reconnecting would mean a decent overhaul to carrot code. msg_id queues being left around might be easy to fix in carrot now that I discover why termie needed to use Consumer.fetch() vs Consumer.wait() to make them work | 04:01 |
| comstud | (the multicall stuff) | 04:01 |
| comstud | but i'd rather leave carrot alone and let it die | 04:03 |
| comstud | switching the default gives the advantage of all of the other non-rpc tests using kombu vs carrot as well | 04:06 |
| vishy | agreed | 04:07 |
| comstud | ok | 04:07 |
| comstud | pushed up default change | 04:07 |
| vishy | nice | 04:07 |
| *** rnorwood has joined #openstack-dev | 04:09 | |
| comstud | thnx for lookin | 04:09 |
| vishy | yw | 04:09 |
| comstud | we might be able to lower some sleeps in these tests at some point | 04:11 |
| comstud | i kludge the 'memory' kombu transport to catch things pretty quickly | 04:11 |
| comstud | seems faster than fake_rabbit w/ carrot | 04:11 |
| comstud | kombu test runs a lot more quickly than the carrot test | 04:12 |
| comstud | test_rpc_carrot: | 04:12 |
| comstud | Ran 10 tests in 0.773s | 04:12 |
| comstud | test_rpc_kombu: | 04:13 |
| comstud | Ran 13 tests in 0.271s | 04:13 |
| johan_-_ | comstud: there was a deeper reason to use fetch instead of consume? | 04:13 |
| comstud | johan: No, I'd suggested termie use .wait() when I reviewed his multicall branch... but.. there were issues with tests and there was a desire to get that in quickly | 04:14 |
| comstud | I suspect i might know what the issue was now.. | 04:15 |
| comstud | though I haven't tried to fix carrot to test my theory | 04:15 |
| johan_-_ | when i switched to using iterconsume(), the code worked fine, but the tests required some deep surgery | 04:16 |
| comstud | yeah | 04:16 |
| johan_-_ | seems like we should hold tests to the same standard of complexity as the rest of the code :/ | 04:16 |
| comstud | I ended up making very minimal changes to tests.. with kombu... and I bet I can make carrot use iterconsume without any issue now | 04:16 |
| comstud | The main thing was the tests needed to consume in a greenthread... but the tests were not monkey patched | 04:17 |
| johan_-_ | ahh | 04:17 |
| comstud | so i think the greenthread was blocking | 04:17 |
| johan_-_ | that could be the cause the problems i ran into | 04:17 |
| johan_-_ | btw, thank you for figuring all of this stuff out :) | 04:18 |
| comstud | i created a consume_in_thread() interface in carrot/kombu now, which handles all of this for you. cleans up service.py a bit as well. | 04:18 |
| comstud | and made sure carrot/kombu monkey_patch things | 04:18 |
| comstud | johan: sure thing | 04:18 |
| comstud | johan: thank you for digging into that initial msg_queue problem :) | 04:19 |
| comstud | i'm still annoyed with the amqp spec, but.. whatever. | 04:20 |
| johan_-_ | yeah, but it's relatively minor | 04:21 |
| johan_-_ | probably more important things to tackle after your branch gets merged | 04:21 |
| comstud | yeah, unless you use 'fetch' or the equiv that does a amqp basic_get() | 04:21 |
| comstud | but | 04:21 |
| comstud | repeat { fetch(); sleep(.1) } is much less efficient anyway | 04:22 |
| comstud | heh. | 04:22 |
| johan_-_ | yeah, i'm not a fan of polling :) | 04:22 |
| comstud | same | 04:22 |
| comstud | it'll be all good now :) | 04:22 |
| johan_-_ | carrot calls sys.exit()? | 04:23 |
| comstud | yeah, when it can't connect to rabbit after x attempts | 04:23 |
| comstud | i duplicated it kombu, but that only gets hit if you set max_retries > 0 now | 04:24 |
| comstud | default is 0 == reconnect forever | 04:24 |
| comstud | fixed carrot to support '0' as forever | 04:24 |
| johan_-_ | when would someone want to set max_retries > 0? | 04:24 |
| comstud | No idea. :) | 04:25 |
| comstud | If someone wants to rip that out, we can do that later | 04:25 |
| comstud | I don't like the sys.exit() either | 04:25 |
| comstud | I can see it being okay *maybe* on daemon startup only.. | 04:26 |
| comstud | but not after the daemon has been running a while | 04:26 |
| johan_-_ | i'm not sure why it would be needed on startup either | 04:27 |
| johan_-_ | but yeah, if it's there already, let another merge fix it | 04:27 |
| johan_-_ | does the DB code give up and sys.exit too? | 04:27 |
| comstud | I found it useful because I'd forget to restart rabbit after a reboot | 04:27 |
| comstud | and I could see quickly what was going on | 04:27 |
| johan_-_ | i see in the logs that rabbitmq wasn't started :) | 04:28 |
| comstud | arguably you would also see it pretty easily looking at logs | 04:28 |
| comstud | :) | 04:28 |
| johan_-_ | then i start it and it's fine | 04:28 |
| comstud | i hear ya. :) | 04:28 |
| comstud | johan: pretty sure 'no' on the DB code | 04:28 |
| comstud | pretty sure it'll just throw an exception | 04:29 |
| comstud | i don't think there's retries either | 04:30 |
| *** nati has joined #openstack-dev | 04:32 | |
| *** novas0x2a|lapto1 has quit IRC | 04:41 | |
| *** zul has quit IRC | 04:49 | |
| *** zul has joined #openstack-dev | 04:50 | |
| *** nati has quit IRC | 05:25 | |
| *** nati has joined #openstack-dev | 05:29 | |
| *** nati has quit IRC | 05:36 | |
| vishy | comstud: I'd like to not have to do greenthread sleeps in in test_cloud to wait for rpcs to finish. It would be nice if we had a "wait for queues to be empty call so that we didn't have to manually throw sleeps in there so casts and calls can finish | 05:48 |
| vishy | comstod: or a fake version of cast/call that just goes ahead and finishes in the same greenthread | 05:49 |
| vishy | comstud: instead of starting a new one | 05:49 |
| *** nati has joined #openstack-dev | 05:58 | |
| *** zul has quit IRC | 05:59 | |
| *** zul has joined #openstack-dev | 06:00 | |
| *** nati has quit IRC | 06:00 | |
| comstud | vishy: hmm | 06:08 |
| comstud | vishy: it's really more than waiting for the event to picked up by the consumer... the consumer also needs to finish all of its work | 06:43 |
| comstud | vishy: which is almost like saying.. 'cast' needs to behave like 'call' in tests. | 06:43 |
| zykes- | how does auth work in glance if there's no use of keystone ? | 06:51 |
| comstud | i think there's no auth w/o keystone | 06:51 |
| comstud | for glance | 06:51 |
| *** reidrac has joined #openstack-dev | 06:55 | |
| zykes- | kk | 07:21 |
| comstud | vishy: very interestingly, changing test_cloud to make rpc.cast do a rpc.call instead... (with a fix to start up a fake ajax_console_proxy)... uncovers a few broken tests. | 07:52 |
| comstud | vishy: which is slightly disconcerting | 07:53 |
| *** adiantum has quit IRC | 07:55 | |
| *** ewindisch has joined #openstack-dev | 07:58 | |
| *** adiantum has joined #openstack-dev | 08:01 | |
| vishy | comstud: interesting. I thought it was already stubbed to use call instead of cast | 08:01 |
| comstud | vishy: it's stubbed to sleep after a cast | 08:01 |
| comstud | vishy: i just pushed up lp:~cbehrens/nova/test-sleep-cleanup as a WIP | 08:01 |
| comstud | branched from rpc-kombu | 08:02 |
| vishy | comstud: actually that doesn't surprise me too much, i know that some of the tests were checking for values that only exist before the cast/call finishes | 08:02 |
| vishy | for example casting to run an instance but checking for scheduling state, etc. | 08:02 |
| comstud | there's some missing arguments when creating volumes and snapshots | 08:02 |
| comstud | in the tests | 08:02 |
| comstud | that cause exceptions | 08:02 |
| vishy | comstud: ha, that could be recent additions, the volume code changed recently | 08:03 |
| comstud | but... somehow the cast doesn't let those cases get it | 08:03 |
| comstud | hit | 08:03 |
| comstud | well | 08:03 |
| comstud | it is | 08:03 |
| comstud | but the tests pass in trunk | 08:03 |
| comstud | which is concerning.. because that means there's code not being covered | 08:03 |
| vishy | comstud: good you found it though. Right but the code was changed and the tests passed, so the author of the changes never fixed them | 08:03 |
| comstud | which i switch to 'call', they show up | 08:03 |
| comstud | correct | 08:04 |
| vishy | they probably were showing exceptions in the logs, if anyone bothered to check | 08:04 |
| comstud | er | 08:05 |
| comstud | the tests were what's broken, not the actual code | 08:05 |
| comstud | so yeah, they didn't update the tests | 08:05 |
| comstud | but | 08:05 |
| comstud | updating the tests wouldn't have made any change in behavior | 08:05 |
| comstud | unless they were working properly..which apparently they aren't | 08:05 |
| comstud | heh | 08:06 |
| comstud | let's see what kind of a speed up this is | 08:07 |
| *** adiantum has quit IRC | 08:07 | |
| comstud | eh, 53 seconds before my changes... 49 after | 08:08 |
| comstud | heh. | 08:08 |
| comstud | i removed pretty much every sleep from test_cloud | 08:08 |
| zykes- | vishy: do you ever sleep ? ;) | 08:09 |
| *** tudamp has joined #openstack-dev | 08:09 | |
| comstud | what is this sleep thing you speak of? | 08:09 |
| zykes- | :p | 08:11 |
| comstud | :) | 08:12 |
| *** adiantum has joined #openstack-dev | 08:12 | |
| vishy | comstud: aww yeah, 4 seconds baby | 08:14 |
| vishy | zykes-: never | 08:14 |
| comstud | vishy: test_cloud changes seem to work with carrot as well as kombu | 08:15 |
| *** jakedahn has quit IRC | 08:15 | |
| zykes- | ;) | 08:16 |
| comstud | vishy: i see test_adminapi and test_abstract_scheduler do similar stuff..i can take a look at fixing those also... later. merge prop soon. | 08:17 |
| comstud | dunno what else might benefit from switching cast to a call.. and what other issues might be uncovered like test_cloud | 08:18 |
| *** adiantum has quit IRC | 08:19 | |
| comstud | i'm gonna go search for this thing called 'sleep' | 08:19 |
| comstud | 'night :) | 08:19 |
| *** adiantum has joined #openstack-dev | 08:24 | |
| *** adiantum has quit IRC | 08:42 | |
| *** darraghb has joined #openstack-dev | 08:46 | |
| *** adiantum has joined #openstack-dev | 08:48 | |
| *** adiantum has quit IRC | 08:55 | |
| *** adiantum has joined #openstack-dev | 09:02 | |
| *** adiantum has quit IRC | 09:06 | |
| *** adiantum has joined #openstack-dev | 09:11 | |
| *** adiantum has quit IRC | 09:18 | |
| *** adiantum has joined #openstack-dev | 09:25 | |
| *** adiantum has quit IRC | 09:33 | |
| *** tsuzuki_ has quit IRC | 09:42 | |
| *** markmc has joined #openstack-dev | 10:10 | |
| *** zigo has joined #openstack-dev | 10:51 | |
| *** markvoelker has joined #openstack-dev | 10:52 | |
| *** mfer has joined #openstack-dev | 11:33 | |
| *** mnot has quit IRC | 11:33 | |
| *** zigo has quit IRC | 11:48 | |
| *** zigo has joined #openstack-dev | 11:49 | |
| *** zigo-_- has joined #openstack-dev | 11:53 | |
| *** zigo has quit IRC | 11:54 | |
| *** zigo-_- has quit IRC | 11:57 | |
| *** bsza has joined #openstack-dev | 12:12 | |
| *** nati has joined #openstack-dev | 12:14 | |
| *** mfer has quit IRC | 12:19 | |
| *** lts has joined #openstack-dev | 12:25 | |
| zykes- | hmmm | 12:29 |
| *** doude has quit IRC | 12:29 | |
| *** doude has joined #openstack-dev | 12:29 | |
| zykes- | wouldn't JSON support within webob.exc be usedful ? | 12:29 |
| *** rnorwood has quit IRC | 12:31 | |
| *** mdomsch has quit IRC | 12:34 | |
| *** rods has joined #openstack-dev | 12:35 | |
| *** jaypipes has quit IRC | 12:39 | |
| *** lorin1 has joined #openstack-dev | 12:51 | |
| *** lorin1 has quit IRC | 12:58 | |
| *** martine_ has joined #openstack-dev | 13:00 | |
| *** sandywalsh has quit IRC | 13:09 | |
| *** mfer has joined #openstack-dev | 13:11 | |
| *** nati has quit IRC | 13:18 | |
| *** sandywalsh has joined #openstack-dev | 13:24 | |
| *** nati has joined #openstack-dev | 13:30 | |
| *** bcwaldon has joined #openstack-dev | 13:32 | |
| *** rnorwood has joined #openstack-dev | 13:47 | |
| *** kbringard has joined #openstack-dev | 13:48 | |
| *** mattray has joined #openstack-dev | 13:53 | |
| *** bsza has quit IRC | 13:54 | |
| *** bsza has joined #openstack-dev | 13:55 | |
| *** nati has quit IRC | 13:56 | |
| *** mattray has quit IRC | 13:57 | |
| *** sandywalsh has quit IRC | 14:01 | |
| *** Binbin has joined #openstack-dev | 14:01 | |
| *** deshantm_laptop has joined #openstack-dev | 14:04 | |
| *** cp16net has joined #openstack-dev | 14:11 | |
| *** sandywalsh has joined #openstack-dev | 14:14 | |
| *** mdomsch has joined #openstack-dev | 14:23 | |
| *** vladimir3p has joined #openstack-dev | 14:35 | |
| *** code_franco has joined #openstack-dev | 14:47 | |
| *** nati has joined #openstack-dev | 14:53 | |
| *** heckj has joined #openstack-dev | 14:53 | |
| *** amccabe has joined #openstack-dev | 14:54 | |
| *** deshantm_laptop has quit IRC | 14:55 | |
| *** jaypipes has joined #openstack-dev | 14:58 | |
| *** reidrac has quit IRC | 15:06 | |
| *** dragondm has joined #openstack-dev | 15:08 | |
| *** deshantm_laptop has joined #openstack-dev | 15:11 | |
| *** bcwaldon has quit IRC | 15:14 | |
| *** bcwaldon has joined #openstack-dev | 15:15 | |
| Vek | can I get more eyes on https://code.launchpad.net/~klmitch/nova/lp832369/+merge/72964 ? | 15:16 |
| *** mattray has joined #openstack-dev | 15:23 | |
| *** mattray has quit IRC | 15:23 | |
| openstackgerrit | A change was merged to openstack/glance: Fix cached-images API endpoint https://review.openstack.org/368 | 15:30 |
| openstackjenkins | Project swift build #314: SUCCESS in 1 min 5 sec: https://jenkins.openstack.org/job/swift/314/ | 15:33 |
| openstackjenkins | Tarmac: Fix for object auditor. It doesn't close files that are quarantined for certain reasons, zero byte files for one, which will cause it to eventually crash due to keeping too many files open. Thanks David Kranz for finding / reporting this!! | 15:33 |
| vladimir3p | Folks, have nybody seen issues with "leaking" MySQL connections? We've started to seen such issue some time ago. Periodically some of our servers complaining about inability to connect to MySQL (too many connections). SHOW PROCESSLIST show tons of connections and all of them are executing Sleep | 15:39 |
| *** misheska has joined #openstack-dev | 15:40 | |
| *** Gordonz has joined #openstack-dev | 15:40 | |
| *** deshantm_laptop has quit IRC | 15:45 | |
| openstackjenkins | Project swift build #315: SUCCESS in 45 sec: https://jenkins.openstack.org/job/swift/315/ | 15:47 |
| openstackjenkins | Tarmac: Proposed fix for bug #837428. | 15:47 |
| uvirtbot | Launchpad bug 837428 in swift "HTTP_X_NEWEST requests require objects to have a timestamp header" [Undecided,New] https://launchpad.net/bugs/837428 | 15:47 |
| openstackjenkins | To test: | 15:47 |
| openstackjenkins | Create an object without an x-timestamp or x-put-timestamp header and request the object with the X-newest header. | 15:47 |
| openstackjenkins | Without the patch, the proxy will return a 404 object not found even though the storage nodes return the object to the proxy. | 15:47 |
| openstackjenkins | With the patch, the proxy will return the first object received unless an object with a newer timestamp header exists. | 15:47 |
| *** nati has quit IRC | 15:56 | |
| *** nati has joined #openstack-dev | 15:57 | |
| *** nati has quit IRC | 16:01 | |
| *** mdomsch has quit IRC | 16:05 | |
| *** ecarlin has joined #openstack-dev | 16:16 | |
| *** ecarlin has quit IRC | 16:17 | |
| *** tudamp has left #openstack-dev | 16:19 | |
| *** mdomsch has joined #openstack-dev | 16:22 | |
| *** nati has joined #openstack-dev | 16:26 | |
| *** jdurgin has joined #openstack-dev | 16:28 | |
| *** nati has quit IRC | 16:41 | |
| *** nati has joined #openstack-dev | 16:41 | |
| *** novas0x2a|laptop has joined #openstack-dev | 16:43 | |
| *** zaitcev has joined #openstack-dev | 16:45 | |
| *** nati has quit IRC | 16:46 | |
| *** nati has joined #openstack-dev | 16:57 | |
| vishy | comstud: sounds like an excellent thing to tackle during cleanup, so go for it. | 16:57 |
| vishy | vladimir3p: that sounds nasty | 16:57 |
| vishy | Vek: btw, it does look like the nova-manage image commands are broken | 16:58 |
| vishy | at least according to the question that Armando posted | 16:58 |
| comstud | vishy: cools | 17:00 |
| tr3buchet | jaypipes: was just reading your testingplan etherpad. for all the continuous deployment etc, what's the plan for testing deploys, functional tests that is | 17:00 |
| *** foods is now known as adam_g | 17:05 | |
| vladimir3p | vishy: I suppose we almost figured this out | 17:12 |
| vladimir3p | vishy: when we start only a CloudController (with all services) it opens ~56 connections | 17:12 |
| vladimir3p | majority of them after that remains in Sleep and counter increasing | 17:13 |
| *** ccorrigan has quit IRC | 17:13 | |
| vladimir3p | for node with volume & compute services only it is about 25-30 connections | 17:13 |
| vladimir3p | which means that we pretty close arrived to max of mySQL connections ... | 17:13 |
| vladimir3p | I bet something changed recently that causes it ... we've not seen it before... | 17:14 |
| vladimir3p | vishy: if you will have a sec try to check # of open connections in your config (SHOW FULL PROCESSLIST ; | 17:15 |
| *** vladimir3p has quit IRC | 17:16 | |
| vishy | vladimiir3p: yes we use connection pooling | 17:17 |
| vishy | vladimir3p: it was a recent change. But it shouldn't just keep adding connecitons over and over. | 17:17 |
| *** vladimir3p has joined #openstack-dev | 17:21 | |
| *** darraghb has quit IRC | 17:22 | |
| *** bcwaldon has quit IRC | 17:32 | |
| *** nati has quit IRC | 17:33 | |
| *** nati has joined #openstack-dev | 17:34 | |
| *** nati has quit IRC | 17:38 | |
| *** Tushar has joined #openstack-dev | 17:42 | |
| Vek | can I get more eyes on https://code.launchpad.net/~klmitch/nova/lp832369/+merge/72964 ? | 17:45 |
| cp16net | question: Would it be ok to use the python _ to ignore returned values from function call even though there is the _() function that provides internationalization? Would this cause confusion? | 17:46 |
| vishy | cp16net: I would suggest _something | 17:48 |
| vishy | instead | 17:49 |
| cp16net | vishy: thats what i was thinking as well just wanted to know what the "standard" if any has been set to on this topic in nova | 17:49 |
| code_franco | if _something why not just something instead | 17:50 |
| cp16net | http://stackoverflow.com/questions/4859909/reason-for-assignment-to/ | 17:51 |
| code_franco | cheater | 17:52 |
| cp16net | this was a suggestion calling it ignored. | 17:52 |
| cp16net | :) | 17:52 |
| *** nati has joined #openstack-dev | 17:53 | |
| *** nickon has joined #openstack-dev | 17:53 | |
| vishy | code_franco: the reason is to make it clear that you are not using the value. Clearer for readers of the code and for automated checkers like pylint | 17:55 |
| vishy | for example user, _sep, project = combined.partition(':') | 17:56 |
| code_franco | You don't think that might cause confusion with the convention to use _names for internal-use-only class methods, though? | 17:58 |
| code_franco | Anyway, it's already been factored out of the code that started the discussion | 17:59 |
| *** joesavak has joined #openstack-dev | 18:00 | |
| *** jsavak has joined #openstack-dev | 18:03 | |
| vishy | blamar: hey can bcwaldon recheck this: https://code.launchpad.net/~cloudbuilders/nova/os-simple-usage/+merge/72499 or can someone overrule the needs fixing? I did the changes he requested | 18:05 |
| *** joesavak has quit IRC | 18:06 | |
| blamar | vishy: talked to him a couple minutes ago he said he was running it through tests and then was going to approve | 18:08 |
| blamar | vishy: lemme find him | 18:08 |
| *** sandywalsh_ has joined #openstack-dev | 18:08 | |
| vishy | cool thx | 18:08 |
| *** sandywalsh has quit IRC | 18:11 | |
| blamar | vishy: approved | 18:14 |
| *** bcwaldon has joined #openstack-dev | 18:14 | |
| vishy | cool thx | 18:14 |
| vishy | comstud: see the note from dprince. Apparently 1.0.4 works and 1.2.1, but we lost 1.1.3 :) | 18:18 |
| sandywalsh_ | where is the wadl file for os api stored? | 18:20 |
| bcwaldon | dont think a wadl exists | 18:20 |
| sandywalsh_ | bcwaldon, ah, rats | 18:20 |
| sandywalsh_ | bcwaldon, is there a published copy of the api spec around? | 18:20 |
| sandywalsh_ | 1.1 | 18:20 |
| bcwaldon | yeah, latest is always at docs.openstack.org | 18:20 |
| sandywalsh_ | thx | 18:20 |
| jaypipes | tr3buchet: sorry, didn't see your note until now... the plan for functional deployment tests is also in that document. We have a Jenkins builder that fires puppet modules that pull the latest PPA packages and runs functional tests against them. The deployment will be configurable (parameterized similar to how smokestack does things), as will the method of dpleoyment (puppet vs. chef, cobbler vs. crowbar, etc). The functional test framework is being c | 18:22 |
| jaypipes | onsolidated currently to include stuff from stacktester and kong, and that will be run against the deployed cluster... more info to follow, gotta run. | 18:22 |
| *** jaypipes has quit IRC | 18:22 | |
| bcwaldon | sandywalsh_: actually, I think I lied to you | 18:22 |
| bcwaldon | sandywalsh_: one sec | 18:22 |
| bcwaldon | sandywalsh_: http://bazaar.launchpad.net/~annegentle/openstack-manuals/trunk/view/head:/doc/source/docbkx/openstack-compute-api-1.1/os-compute-1.1.wadl | 18:24 |
| sandywalsh_ | bcwaldon, nice! thx | 18:24 |
| sandywalsh_ | bcwaldon, surprised that's not in /nova ? | 18:24 |
| bcwaldon | sandywalsh_: no problem you're lucky I happened to be in the code already | 18:24 |
| annegentle | sandywalsh_: bcwaldon: I'll be moving the wadl etc. to a nova-api repo | 18:26 |
| bcwaldon | annegentle: sounds great | 18:26 |
| comstud | vishy: oh gosh.. ok. i'll check out 1.1.3 | 18:26 |
| bcwaldon | annegentle: real quick, I'm trying to build the docs. Gimme a pointer? | 18:27 |
| sandywalsh_ | annegentle, thx! | 18:27 |
| bcwaldon | annegentle: not sure where to run mvn | 18:27 |
| *** bengrue has joined #openstack-dev | 18:28 | |
| comstud | hm, test_rpc_kombu passes with 1.1.3 for me | 18:29 |
| comstud | *looks at the note* | 18:29 |
| annegentle | bcwaldon: oh sure, run it in the directory that contains the pom.xml. For openstack-manuals, run it in /doc/ | 18:30 |
| annegentle | bcwaldon: if that doesn't work, I may need to send you .m2 settings.xml file | 18:30 |
| *** SpamapS has quit IRC | 18:44 | |
| *** AhmedSoliman has joined #openstack-dev | 18:45 | |
| *** SpamapS has joined #openstack-dev | 18:52 | |
| *** SpamapS has joined #openstack-dev | 18:52 | |
| *** mattray has joined #openstack-dev | 18:53 | |
| *** nati has quit IRC | 18:54 | |
| comstud | vishy: memory transport bug in 1.1.3 :-/ i've kludged something in and tests pass with 1.1.3 now.. Now I need to see if this breaks 1.0.4 and/or 1.2.1 | 18:57 |
| *** mdomsch has quit IRC | 19:06 | |
| *** heckj has quit IRC | 19:08 | |
| *** mdomsch has joined #openstack-dev | 19:12 | |
| *** mwhooker has joined #openstack-dev | 19:23 | |
| openstackjenkins | Project swift build #316: SUCCESS in 30 sec: https://jenkins.openstack.org/job/swift/316/ | 19:32 |
| openstackjenkins | Tarmac: Fixes multi node docs to point to the right ppa | 19:32 |
| *** markmc has quit IRC | 19:36 | |
| *** AhmedSoliman has quit IRC | 19:41 | |
| *** nickon has quit IRC | 19:45 | |
| vishy | bcwaldon: ping | 19:52 |
| bcwaldon | vishy: yes | 19:52 |
| vishy | bcwaldon: https://code.launchpad.net/~cloudbuilders/nova/os-keypair-integration/+merge/72140 | 19:53 |
| vishy | so it wasn't a super issue | 19:53 |
| vishy | I fixed it without changing and pushed. | 19:53 |
| bcwaldon | vishy: okay, looking | 19:53 |
| vishy | The question is do you want _get_default_keypair? | 19:54 |
| vishy | i have a second version that does it that way | 19:54 |
| vishy | bcwaldon: ^^ | 19:54 |
| bcwaldon | vishy: hold on one sec, looking at the test failure | 19:54 |
| bcwaldon | vishy: from tarmac, not you | 19:55 |
| vishy | bcwaldon: other version http://pastie.org/2461750 | 20:00 |
| bcwaldon | vishy: ok, still trying to nail down these test failures | 20:01 |
| vishy | np | 20:01 |
| vishy | going to lunch anyway, just let me know which you prefer | 20:01 |
| bcwaldon | vishy: kk, I'll get back to you | 20:01 |
| openstackgerrit | James E. Blair proposed a change to openstack/openstack-ci: Sync openstack-deploy-core group to gerrit. https://review.openstack.org/369 | 20:03 |
| openstackgerrit | A change was merged to openstack/openstack-ci: Sync openstack-deploy-core group to gerrit. https://review.openstack.org/369 | 20:04 |
| comstud | well shoot | 20:15 |
| *** jaypipes has joined #openstack-dev | 20:18 | |
| comstud | mtaylor, soren: have a merge failure i need some help with. https://code.launchpad.net/~cbehrens/nova/rpc-kombu/+merge/73096 | 20:18 |
| comstud | mtaylor, soren: appears kombu could not load.. | 20:18 |
| soren | comstud: Erk. | 20:20 |
| comstud | soren: the branch has been tested with the ppa version, so.. | 20:22 |
| comstud | soren: I'm not sure what gives | 20:22 |
| soren | comstud: Let me see if I can find out. | 20:24 |
| soren | Gah, I don't even know where these things run nowadays. | 20:24 |
| comstud | unfortunately import_object eats the exception | 20:24 |
| soren | I hate that thing. | 20:24 |
| soren | For that very reason. | 20:24 |
| comstud | soren: Yea, wasn't sure if you could help or if it's strictly a monty thing | 20:24 |
| openstackgerrit | Rick Harris proposed a change to openstack/glance: Adding Keystone support for Glance client. https://review.openstack.org/350 | 20:25 |
| openstackgerrit | James E. Blair proposed a change to openstack/openstack-chef: Add rfc.sh to support git review. https://review.openstack.org/370 | 20:25 |
| soren | comstud: soren@nova:~$ dpkg -l python-kombu | 20:26 |
| soren | No packages found matching python-kombu. | 20:26 |
| soren | soren@nova:~$ dpkg -l python-kombu | 20:26 |
| soren | No packages found matching python-kombu. | 20:26 |
| soren | comstud: That would explain it. | 20:26 |
| comstud | hrm | 20:26 |
| comstud | how do tests pass with glance having a dependency on kombu then? | 20:27 |
| soren | They proably run somewhere else. | 20:27 |
| comstud | well, nova has a dependency on glance, so.. | 20:28 |
| *** mdomsch has quit IRC | 20:28 | |
| soren | comstud: /me fixes | 20:28 |
| comstud | soren: thank you, sir | 20:28 |
| openstackgerrit | A change was merged to openstack/openstack-chef: Add rfc.sh to support git review. https://review.openstack.org/370 | 20:28 |
| soren | comstud: Then monty can kill me on Monday if this is the wrong way to do it. | 20:28 |
| soren | (He's at Burning Man this week) | 20:29 |
| soren | comstud: There. Should be all good now. | 20:29 |
| comstud | Burning Man... REALLY? | 20:29 |
| comstud | i'm disappointed. | 20:29 |
| comstud | soren: I've got your back | 20:29 |
| comstud | :) | 20:29 |
| *** nati has joined #openstack-dev | 20:29 | |
| jeblair | soren: what did you do to fix? | 20:29 |
| comstud | apt-get install python-kombu i assume | 20:29 |
| comstud | :) | 20:29 |
| soren | jeblair: Installed python-kombu on the "nova" box. | 20:29 |
| soren | jeblair: Yeah, what comstud said. | 20:30 |
| jeblair | is python-kombu a build-dep for nova? | 20:30 |
| soren | It is now. | 20:30 |
| soren | Well... | 20:30 |
| jeblair | adding it as a build-dep to the package will cause the nova boxes to install it automatically | 20:30 |
| jeblair | fixing it with apt-get is fine | 20:30 |
| Vek | can I get more eyes on https://code.launchpad.net/~klmitch/nova/lp832369/+merge/72964 ? | 20:30 |
| soren | Right, but it's not a build dep until this merge prop is done. | 20:31 |
| soren | ...and this merge prop can't finish until it's installed. Catch 22. | 20:31 |
| jeblair | yes, there is sometimes a catch-22. | 20:31 |
| jeblair | (sometimes other props get through first and break the lock) | 20:31 |
| jeblair | but hey, as long as it gets into the package as a build dep, future nova builders will have it installed | 20:31 |
| comstud | i'm gonna send a note to the list about this new dependency | 20:31 |
| jeblair | so that's what really matters | 20:31 |
| comstud | we'd thought this was already a dependency, so.. | 20:32 |
| soren | Perhaps glance doesn't state it as a dependency or something. | 20:33 |
| soren | DUnno. | 20:33 |
| *** nati has quit IRC | 20:34 | |
| comstud | it would make sense that if you only use the client, it doesn't depend on kombu | 20:36 |
| comstud | (glance client) | 20:36 |
| Vek | kombu is in glance's pip-requires | 20:37 |
| comstud | Yeah | 20:37 |
| comstud | but | 20:37 |
| comstud | i dunno :) | 20:37 |
| jeblair | soren, comstud: python-kombu should be listed as a build-depends in the control file for nova in the nova packaging | 20:37 |
| jeblair | http://bazaar.launchpad.net/~openstack-ubuntu-packagers/nova/ubuntu/view/head:/debian/control | 20:37 |
| Daviey | vishy: Hmm, kombu branch is landing in diablo? | 20:38 |
| comstud | ok.. is that something I should add and merge prop then? | 20:38 |
| Daviey | I thought it was agreed it was being put on hold? | 20:38 |
| *** med_out is now known as medberry | 20:38 | |
| comstud | Daviey: I'd not seen anything about putting it on hold if it still had carrot support.. | 20:39 |
| soren | Daviey: PTL decree: https://lists.launchpad.net/openstack/msg03833.html | 20:39 |
| Daviey | comstud: Have you tested against 1.0.4 ? | 20:39 |
| comstud | Daviey: Yeppers | 20:39 |
| Daviey | comstud: and it works fine on that version? | 20:39 |
| *** jsavak has quit IRC | 20:39 | |
| comstud | Daviey: Yep | 20:39 |
| Daviey | comstud: that is GREAT news. | 20:40 |
| comstud | Daviey: 1.0.4, 1.1.3, 1.2.1.. you name it | 20:40 |
| soren | Daviey: That's what I installed on the tarmac box, so that's at least what the tests are run against. | 20:40 |
| comstud | hopefully various versions in the middle too | 20:40 |
| * Daviey cries at the work that has gone into making carrot good for this release, based on the outcome of last weeks meeting :) | 20:40 | |
| * soren takes python-carrot out back and shoots it in the head | 20:40 | |
| soren | There. Sorted. | 20:40 |
| comstud | python-carrot is still required | 20:42 |
| comstud | for now | 20:43 |
| Daviey | comstud: Thanks for pushing that, really appreciate it! | 20:43 |
| Daviey | comstud: oh? | 20:43 |
| comstud | for tests | 20:43 |
| comstud | and a fallback solution in case kombu breaks | 20:43 |
| comstud | carrot will get ripped out post-diablo | 20:43 |
| comstud | IMO | 20:43 |
| Daviey | comstud: the test cases still use the carrot functions? | 20:43 |
| comstud | Daviey: tests will test both carrot and kombu | 20:44 |
| Daviey | can that not be an either/or? | 20:44 |
| comstud | sure: rm nova/tests/test_rpc_carrot.py | 20:44 |
| comstud | :) | 20:44 |
| comstud | So, no, not really.. | 20:44 |
| Daviey | comstud: I mean, if carrot import fails SKIP test? | 20:44 |
| comstud | we want carrot to still function in diablo just in case | 20:44 |
| *** joesavak has joined #openstack-dev | 20:44 | |
| comstud | so it should really have tests | 20:45 |
| soren | Daviey: You could propse a patch that only tests the carrot things if carrot is available. | 20:45 |
| Daviey | soren: ack | 20:45 |
| comstud | yeah | 20:45 |
| comstud | that should be an easy one after this merges | 20:45 |
| comstud | I'd rather not make any additional changes if I don't have to right now | 20:45 |
| comstud | to make sure this gets in | 20:45 |
| soren | Daviey: Or add a --im-convinced-I-wont-need-carrot flag that skips them. | 20:45 |
| Daviey | comstud: i assume you saw the merge failed? | 20:45 |
| soren | What, again? | 20:45 |
| comstud | Daviey: Yes, due to python-kombu not installed | 20:45 |
| Daviey | ahh | 20:46 |
| comstud | Daviey: soren just fixed that | 20:46 |
| comstud | i'm waiting for Approves again | 20:46 |
| * soren clicked apporved. | 20:46 | |
| soren | approved, even. | 20:46 |
| comstud | ok | 20:46 |
| comstud | thnx :) | 20:46 |
| soren | Just now. | 20:46 |
| soren | np :) | 20:46 |
| comstud | hehe | 20:46 |
| soren | Thanks for doing this. I'm happy to see this land. | 20:47 |
| comstud | Sure, no problem | 20:47 |
| soren | Mmm.... More tests for libvirt are brewing. | 20:53 |
| comstud | i've got some test fixes coming, also | 20:59 |
| bcwaldon | vishy: I'm more for the version in the paste you sent me | 20:59 |
| bcwaldon | vishy: but I don't want to give the ability for someone to provide key_name in v1.0 | 21:01 |
| Daviey | comstud: Are you going to handle conditional test for carrot / kombu, or should i do that? | 21:01 |
| bcwaldon | vishy: and I think we should just implement _get_default_key for each versioned controller (you'll have to pass in the request body) | 21:01 |
| *** joesavak has quit IRC | 21:01 | |
| comstud | Daviey: I've got some other work, so if you want it... go for it. | 21:03 |
| Daviey | comstud: great stuff, thanks again | 21:03 |
| comstud | Daviey: np | 21:03 |
| medberry | ah, comstud is Chris Behrens. It's all making sense now. | 21:03 |
| comstud | that's me | 21:04 |
| comstud | :) | 21:04 |
| *** zaitcev has left #openstack-dev | 21:05 | |
| comstud | unfortunately i'm stuck with this nickname | 21:05 |
| openstackjenkins | Project nova build #1,313: SUCCESS in 3 min 39 sec: https://jenkins.openstack.org/job/nova/1313/ | 21:10 |
| openstackjenkins | Tarmac: Disassociated previously associated floating ips when calling network_api.associate_floating_ip. Also guard against double-association in the network.manager. | 21:10 |
| *** bcwaldon has quit IRC | 21:15 | |
| openstackjenkins | Project nova build #1,314: SUCCESS in 3 min 40 sec: https://jenkins.openstack.org/job/nova/1314/ | 21:20 |
| openstackjenkins | Tarmac: Implements lp:798876 which is 'switch carrot to kombu'. Leaves carrot as the default for now... decision will be made later to switch the default to kombu after further testing. There's a lot of code duplication between carrot and kombu, but I left it that way in preparation for ripping carrot out later and to keep minimal changes to carrot. | 21:20 |
| openstackjenkins | This also fixes bug: lp:794627 (re-establish connections to carrot when it restarts), but only fixes it in kombu. | 21:20 |
| openstackjenkins | This also fixes bug: lp:803168 (msg-id response queues being left around), but also only fixes it in kombu. | 21:20 |
| openstackjenkins | See those bugs for comments. | 21:20 |
| comstud | woot. | 21:26 |
| Vek | can I get a core to look at https://code.launchpad.net/~klmitch/nova/lp832369/+merge/72964 ? Just need one more approve. | 21:26 |
| comstud | shoot, i forgot to update the description that says 'leaves carrot as default'.. since we decided to make kombu the default. oh well | 21:27 |
| medberry | oh so kombu is the default and it falls back to carrot if no kombu? | 21:28 |
| comstud | no, you'll actually need kombu installed to run tests | 21:29 |
| comstud | and if you want to use carrot, you'll have to set: --rpc_backend=nova.rpc.impl_carrot | 21:30 |
| comstud | to override the default | 21:30 |
| comstud | if someone wants to submit patches that test for carrot/kombu and pick a default based on that... go for it | 21:31 |
| comstud | I've got to move on to some other work | 21:31 |
| *** lts has quit IRC | 21:32 | |
| *** jaypipes has quit IRC | 21:33 | |
| Daviey | bug #838386 | 21:35 |
| uvirtbot | Launchpad bug 838386 in nova "Test suite requires kombu and carrot to be installed" [Undecided,New] https://launchpad.net/bugs/838386 | 21:35 |
| *** cp16net has quit IRC | 21:38 | |
| *** bengrue has quit IRC | 21:44 | |
| *** bengrue has joined #openstack-dev | 21:44 | |
| vishy | bcwaldon: not sure i understand | 21:47 |
| westmaas | vishy: you mind making sure blamar addressed your comments on instance states? https://code.launchpad.net/~rackspace-titan/nova/instance_states/+merge/72502 | 21:48 |
| vishy | bcwaldon: so if someone passes key_name in 1.0, you want it to come back as none? | 21:48 |
| westmaas | vishy: bcwaldon disconnected | 21:49 |
| vishy | westmaas: done and in | 21:49 |
| westmaas | can pass on your question tho | 21:49 |
| westmaas | thank you sir | 21:49 |
| vishy | westmaas: thanks, I'm wondering if it should raise an exception or just not use it | 21:50 |
| vishy | (if a key_name is passed in 1.0 | 21:50 |
| *** kbringard has quit IRC | 21:57 | |
| *** mfer has quit IRC | 22:00 | |
| *** code_franco has quit IRC | 22:06 | |
| *** code_franco has joined #openstack-dev | 22:07 | |
| *** misheska has quit IRC | 22:09 | |
| mwhooker | is christopher macgown around? | 22:12 |
| _0x44 | Yes | 22:14 |
| _0x44 | What's up? | 22:14 |
| mwhooker | I want to run an idea by you for my validator path before I resubmit | 22:14 |
| _0x44 | Oh cool | 22:15 |
| mwhooker | s/path/patch/ | 22:15 |
| _0x44 | While you're at it, you should pull in a userdata validator too | 22:15 |
| _0x44 | Hold on, let me find the URL for the openstack API fix | 22:15 |
| mwhooker | what I'm really trying to do is validate a url path. checking for ascii, or if all characters are printable doesn't really achieve that. I've got a gnarly regex which validates the path per the RFC, and I want to see what you think | 22:16 |
| mwhooker | okay | 22:16 |
| _0x44 | mwhooker: https://code.launchpad.net/~tpatil/nova/bug837534/+merge/73456 | 22:17 |
| _0x44 | mwhooker: Can you paste it on http://paste.openstack.org/? | 22:17 |
| mwhooker | http://pastebin.com/6hKXuwKA | 22:18 |
| mwhooker | okay | 22:18 |
| mwhooker | http://paste.openstack.org/show/2327/ | 22:18 |
| mwhooker | I'm borrowing the regex from http://jmrware.com/articles/2009/uri_regexp/URI_regex.html | 22:18 |
| _0x44 | Sorry, I didn't realize you'd already pasted it somewhere | 22:18 |
| mwhooker | np | 22:19 |
| *** rnorwood has quit IRC | 22:19 | |
| mwhooker | I'll poke at the user data patch | 22:19 |
| *** code_franco has quit IRC | 22:21 | |
| *** jaypipes has joined #openstack-dev | 22:23 | |
| *** medberry is now known as med_out | 22:23 | |
| _0x44 | mwhooker: Looking now, trying to see if there's a cleaner/simpler way | 22:26 |
| mwhooker | would love it if there is | 22:27 |
| openstackgerrit | Yogeshwar Srikrishnan proposed a change to openstack/keystone: Fixing existing wadl.Completing wadl for extension OS-KSADM. https://review.openstack.org/371 | 22:27 |
| mwhooker | I'm incredibly surprised I can't find anything in the python standard library which does this | 22:27 |
| soren | Man, our test suite takes an astounding amount of time. | 22:27 |
| _0x44 | mwhooker: If that was a URL instead of a path fragment, urlparse.urlparse would work and validate everything. | 22:27 |
| soren | mwhooker: Sorry, does what? | 22:27 |
| mwhooker | soren: validates URLs with RFC3986's grammar | 22:28 |
| mwhooker | _0x44: even that doesn't look at the grammar | 22:28 |
| mwhooker | just uses the simple regex in rfc3986 appendix B | 22:28 |
| mwhooker | http://tools.ietf.org/html/rfc3986#appendix-B | 22:30 |
| Vek | Yo, core devs! Please look at https://code.launchpad.net/~klmitch/nova/lp832369/+merge/72964 ... Just need one more approve. (It's short...) | 22:31 |
| _0x44 | mwhooker: Why do you need more than that? Once the urlparse has been parsed into URL components, the path is split from the parameters and netloc | 22:31 |
| mwhooker | if anyone thinks url paths don't need to be validated to this extent (i.e. let a lower layer handle it), please speak up | 22:31 |
| * vishy grants himself the power to ninja-approve patches | 22:32 | |
| vishy | soren: i know, it is depressingly slow :( | 22:32 |
| mwhooker | _0x44: in keeping with the spirit of the patch, I want to make sure strings like "/bad\n/ªpath/" are rejected | 22:32 |
| vishy | switch to dtest, test using all cores! right Vek? | 22:32 |
| *** amccabe has quit IRC | 22:33 | |
| mwhooker | http://paste.openstack.org/show/2328/ | 22:34 |
| creiht | mwhooker: shouldn't the path be encoded properly before you try to parse it? | 22:34 |
| mwhooker | creiht: it's user input | 22:35 |
| mwhooker | I'm trying to validate it | 22:35 |
| mwhooker | if it's escaped: fine. if not, reject it | 22:35 |
| Vek | vishy: heh :) | 22:35 |
| *** negronjl has quit IRC | 22:36 | |
| soren | vishy: I'd much rather get sqlite out of the equation. | 22:37 |
| soren | vishy: Or at least get a file-backed sqlite out of the equation. | 22:37 |
| *** amccabe has joined #openstack-dev | 22:38 | |
| vishy | vishy: yes please | 22:38 |
| vishy | i' mean | 22:39 |
| vishy | soren: yes pleas | 22:39 |
| vishy | e | 22:39 |
| westmaas | soren: you have access to jenkins? | 22:39 |
| westmaas | seems to be stuck | 22:39 |
| westmaas | mtaylor: you around? | 22:39 |
| soren | vishy: eatmydata helps some, but it's still way slower than it ought to be. | 22:40 |
| vishy | soren: we just need to mock out the db calls | 22:40 |
| soren | westmaas: mtaylor's gone for the week. | 22:40 |
| soren | vishy: Yeah. Walk in the part. | 22:40 |
| soren | park, even. | 22:40 |
| westmaas | doh | 22:40 |
| soren | westmaas: I can look, though. | 22:40 |
| soren | westmaas: As can jeblair. | 22:40 |
| vishy | soren: only some tests are still using real db calls | 22:40 |
| soren | vishy: orly? | 22:41 |
| westmaas | soren: cool thanks. Looks stuck on this guy: ~skoubli/nova/lp834633 | 22:41 |
| vishy | soren: so I don't know how much speedup we'll get. For example, none of the openstack api tests are using them | 22:41 |
| westmaas | maybe just set to needs fixing and reset the job | 22:41 |
| soren | westmaas: How is it "stuck"? | 22:43 |
| *** negronjl has joined #openstack-dev | 22:43 | |
| _0x44 | mwhooker: It looks like multi-byte characters get escaped by python anyway: | 22:43 |
| _0x44 | mwhooker: http://paste.openstack.org/show/2329/ | 22:44 |
| *** amccabe has quit IRC | 22:44 | |
| mwhooker | _0x44: that's just how python represents the string | 22:44 |
| westmaas | soren: hm, actually I see lots of failed builds, I thought it was just spinning on that last build for a while, so I'm not so sure | 22:44 |
| _0x44 | mwhooker: Yes, but the only character that would get caught by your REGEX is '\' | 22:45 |
| *** vladimir3p_ has joined #openstack-dev | 22:46 | |
| _0x44 | mwhooker: But you're allowing it in your pchar | 22:46 |
| mwhooker | let me see | 22:46 |
| mwhooker | \n gets caught | 22:47 |
| mwhooker | let me try to stick some unicode in there | 22:47 |
| *** vladimir3p has quit IRC | 22:48 | |
| _0x44 | mwhooker: Actually, because of how you're doing your match it catches it. | 22:48 |
| *** mattray has quit IRC | 22:48 | |
| mwhooker | u"http://fake/bad/path\uFFAA/" gets caught. having trouble with the other glyph in my editor | 22:48 |
| _0x44 | mwhooker: So the only thing I'd suggest for your patch is to precompile it with PATH_VALIDATOR_RE = re.compile(path) and then use PATH_VALIDATOR_RE.match instead | 22:49 |
| mwhooker | _0x44: good idea | 22:49 |
| _0x44 | Also please document that with a link to where you got the regex so that when people come back and look at it their eyes only bleed a little bit. | 22:49 |
| _0x44 | :) | 22:50 |
| soren | vishy: Have you every tried profiling a test run? | 22:57 |
| *** lorin1 has joined #openstack-dev | 22:58 | |
| vishy | soren: I have not, have you? | 22:59 |
| soren | vishy: Yes, and failed :) | 22:59 |
| vishy | that means our tests are sexy, right? | 22:59 |
| openstackjenkins | Project nova build #1,315: SUCCESS in 3 min 39 sec: https://jenkins.openstack.org/job/nova/1315/ | 23:00 |
| openstackjenkins | Tarmac: Fixed and improved the way instance "states" are set. Instead of relying on solely the power_state of a VM, there are now explicitly defined VM states and VM task states which respectively define the current state of the VM and the task which is currently being performed by the VM. | 23:00 |
| *** jaypipes has quit IRC | 23:00 | |
| soren | vishy: Erm... :) | 23:00 |
| creiht | can't you run nosetests to tell you how long each test took, then look for the outliers? | 23:01 |
| vishy | creiht: that will save some time, but it is an overall slowness | 23:01 |
| vishy | we have > 1000 tests | 23:02 |
| *** jaypipes has joined #openstack-dev | 23:02 | |
| *** lorin1 has quit IRC | 23:02 | |
| *** bsza has quit IRC | 23:10 | |
| pvo | soren: I put the tarmac trace in the MP that westmaas pointed out | 23:13 |
| soren | westmaas, pvo: Sorry, guys. Way too tired to do anything useful about this now. :( | 23:15 |
| openstackgerrit | Ziad Sawalha proposed a change to openstack/keystone: Consolidating xsds. Splitting contrib to admin and service. https://review.openstack.org/363 | 23:15 |
| pvo | soren: I think we're ok for now | 23:16 |
| soren | \o/ | 23:16 |
| pvo | we're building. I'll try to peek at that patch in a bit | 23:16 |
| pvo | to find out why is special about it | 23:16 |
| soren | vishy: I got profile stats, btw. | 23:16 |
| vishy | whoot | 23:18 |
| openstackgerrit | A change was merged to openstack/keystone: Consolidating xsds. Splitting contrib to admin and service. https://review.openstack.org/363 | 23:20 |
| *** zns has joined #openstack-dev | 23:20 | |
| vishy | wow we are up to 1700 tests, crazy | 23:21 |
| soren | vishy: ...and I have a bunch more coming soon. | 23:22 |
| vishy | soren: looking forward to it | 23:22 |
| soren | https://code.launchpad.net/~soren/nova/virt-test-improvements | 23:23 |
| soren | I want to stare at it for another day or so first. | 23:23 |
| soren | ...but it's looking really good. | 23:23 |
| vishy | nice | 23:23 |
| soren | My profiling reveals a *lot* of time spent in sqlalchemy (and below). | 23:23 |
| vishy | soren: what is your opinion on the libvirt/xen image handling | 23:24 |
| soren | vishy: A flag is probably a good idea at this stage. | 23:24 |
| soren | ...and then we can discuss at the summit if we want to unify or whatever. | 23:24 |
| vishy | soren: my thought is to switch to the resize method, and just depend on the guest to resize the internal filesystem al a cloud-init | 23:24 |
| pvo | soren: for the keystone import, was thinking we could write out something similar to sampledata.py? | 23:25 |
| pvo | sorry, not soren. vishy ^^ | 23:25 |
| vishy | pvo: yes, sure | 23:25 |
| pvo | cool | 23:25 |
| vishy | pvo: from nova.auth import manager | 23:26 |
| vishy | am = manager.AuthManager() | 23:26 |
| soren | vishy: I think that's reasonable, but I imagine Rackspace will want to do whatever they're used to doing. Hence a flag. | 23:26 |
| vishy | for user in am.get_users(): # add to keystone | 23:26 |
| pvo | vishy: righton | 23:26 |
| vishy | for project in am.get_projects(): # add to keystone | 23:26 |
| vishy | the tricky part is roles | 23:26 |
| soren | Good grief, what am I still doing up? | 23:26 |
| vishy | and access/secret keys | 23:27 |
| soren | Good night, guys. | 23:27 |
| vishy | soren: yes, flag is useful | 23:27 |
| vishy | soren: i think that is what rs is doing though | 23:27 |
| vishy | more or less anyway | 23:27 |
| pvo | vishy: we're going to get the BP out soon for the resize | 23:27 |
| pvo | if it needs more explanation | 23:28 |
| vishy | pvo: I think your standard resize method is a default | 23:28 |
| vishy | * sane default | 23:28 |
| pvo | auto expand the disk and the partition? | 23:28 |
| zns | vishy: I didn't understand the get_usedrs/projects comments above. Is that something you need done in keystone in nova authmanager? | 23:29 |
| vishy | pvo: we can have a flag for using the second disk attach. Implementing in kvm will be annoying for two reasons 1) caching of cow images and 2) block_device_mapping | 23:29 |
| vishy | but i think it is usable | 23:29 |
| vishy | zns: no that is for a migration script from nova to keystone | 23:29 |
| zns | Ah - OK. | 23:29 |
| zns | BTW, sampledata.py calls keystone-manage. It's slow because it spins up keystone each line that runs. We have an optimization for that, so don't worry about optimizing... | 23:30 |
| pvo | zns: yea, I expect it to be slow but it should just be a one time thing for folks | 23:31 |
| *** mwhooker has quit IRC | 23:36 | |
| vishy | zns: btw, has that tools.tracer issue been addressed? | 23:37 |
| vishy | someone was running into it the other day, but I wasn't sure if it was because they were using an older build | 23:37 |
| zns | vishy: yes. tracer.py has been moved under keystone so it's keystone.tools.tracer now. | 23:39 |
| *** zns has quit IRC | 23:39 | |
| vishy | zns: coolness | 23:41 |
| vishy | blamar: ping on https://code.launchpad.net/~vishvananda/nova/lp831999/+merge/72633 | 23:43 |
| *** novas0x2a|laptop has quit IRC | 23:50 | |
| openstackjenkins | Project nova build #1,316: SUCCESS in 3 min 32 sec: https://jenkins.openstack.org/job/nova/1316/ | 23:55 |
| openstackjenkins | Tarmac: Validates that user-data is b64 encoded. | 23:55 |
| *** bengrue has quit IRC | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!