| *** dragondm has quit IRC | 00:26 | |
| *** mszilagyi has quit IRC | 00:51 | |
| *** jdurgin has quit IRC | 01:05 | |
| *** cloudgroups has joined #openstack-dev | 01:47 | |
| *** Binbin has joined #openstack-dev | 03:13 | |
| *** blamar__ has joined #openstack-dev | 03:23 | |
| *** Zangetsue has joined #openstack-dev | 03:24 | |
| *** cloudgroups has left #openstack-dev | 03:48 | |
| *** blamar__ has quit IRC | 04:54 | |
| *** zaitcev has quit IRC | 06:07 | |
| *** Binbin has quit IRC | 06:19 | |
| *** Binbin has joined #openstack-dev | 06:35 | |
| *** jaypipes has joined #openstack-dev | 08:03 | |
| soren | gholt: Excellent, thanks for clearing that up. | 08:51 |
|---|---|---|
| *** Binbin has quit IRC | 09:57 | |
| *** cweidenkeller has quit IRC | 10:22 | |
| *** markvoelker has joined #openstack-dev | 11:18 | |
| *** cweidenkeller has joined #openstack-dev | 11:39 | |
| *** cweidenkeller has quit IRC | 11:56 | |
| *** cweidenkeller has joined #openstack-dev | 11:56 | |
| *** cweidenkeller has quit IRC | 12:14 | |
| *** cweidenkeller has joined #openstack-dev | 12:14 | |
| *** cweidenkeller has quit IRC | 12:38 | |
| *** cweidenkeller has joined #openstack-dev | 12:38 | |
| *** cweidenkeller has quit IRC | 12:47 | |
| *** cweidenkeller has joined #openstack-dev | 12:47 | |
| *** cweidenkeller has quit IRC | 12:51 | |
| *** cweidenkeller has joined #openstack-dev | 12:51 | |
| dabo | I just grabbed the latest trunk, and getting the following error running the unit tests: DuplicateFlagError: The flag 'instances_path' is defined twice. First from nova.compute.manager, Second from nova.compute.manager | 12:53 |
| dabo | Anyone else getting this? Which definition is the correct one? | 12:53 |
| *** dprince has joined #openstack-dev | 12:53 | |
| soren | nova.compute.manager for sure. | 12:54 |
| dabo | soren: that's what I thought, but the other looks like a more recent change. | 12:55 |
| dabo | surprised that both made it into trunk | 12:55 |
| soren | Ok, now I'm confused. | 12:56 |
| * soren stares at "nova.compute.manager" and "nova.compute.manager" | 12:56 | |
| soren | dabo: Are you pulling my leg? | 12:58 |
| dabo | not intentionally :) | 12:58 |
| *** alex-meade has joined #openstack-dev | 12:58 | |
| dabo | that's what the output said. The actual conflict is in nova.scheduler.driver | 12:59 |
| dabo | didnt' realize that the output was messed up | 12:59 |
| dabo | Guess that's a bug in the flags module | 13:00 |
| soren | dabo: Where/when do you get that error? | 13:02 |
| dabo | soren: I ran bzr pull on trunk, and then run_tests.sh | 13:02 |
| soren | dabo: And then you got it right away? | 13:03 |
| dabo | I get a bunch of test fails due to Duplicate Flag error. I was going to trash my trunk and start fresh, but wanted to check here first | 13:04 |
| soren | Gimme a sec. | 13:04 |
| soren | I don't see those errors. | 13:07 |
| dabo | ok, then somehow my trunk is hosed. Guess it's time to wipe and re-branch | 13:07 |
| soren | bzr clean-tree? | 13:07 |
| dabo | Odd how that can happen, since I never edit trunk OMM | 13:08 |
| soren | "bzr clean-tree --unknown --ignored" even. | 13:08 |
| dabo | soren: ok, that cleaned up the problem. | 13:13 |
| dabo | soren: DVCS whine template: <DVCS I use infrequently> sucks! I just want to <task that's no easier in my fave DVCS but I've learned how it works>! | 13:13 |
| soren | What's the equivalent in git? | 13:14 |
| *** mattray has joined #openstack-dev | 13:15 | |
| soren | git clean -x, apparantly. | 13:15 |
| dabo | soren: yeah. But I've never seen a virgin branch go "bad" like this. | 13:15 |
| soren | It's probably due to leftover .pyc files. | 13:15 |
| soren | bzr clean-tree doesn't touch anything other than unknown (unversioned) or ignored (e.g. .pyc) files. | 13:16 |
| soren | This wasn't a $VCS problem. $VCS was just a tool for cleaning up. | 13:17 |
| dabo | it might be something in the venv, which the clean-tree command removed | 13:19 |
| dabo | soren: could be. It's just odd that the only thing I ever do in trunk is pull from lp, and run tests. | 13:23 |
| *** salv-orlando has joined #openstack-dev | 13:24 | |
| *** taturiello has joined #openstack-dev | 13:25 | |
| soren | It's not, really :) | 13:25 |
| soren | Doing that will create all the .pyc files needed to screw something up. | 13:25 |
| *** taturiello has left #openstack-dev | 13:26 | |
| soren | Anyways, /me wanders off for a number of hours. | 13:26 |
| *** bcwaldon has joined #openstack-dev | 13:26 | |
| *** Zangetsue has quit IRC | 13:30 | |
| *** Zangetsue has joined #openstack-dev | 13:31 | |
| *** Zangetsue has quit IRC | 13:57 | |
| *** jkoelker has joined #openstack-dev | 14:06 | |
| *** dragondm has joined #openstack-dev | 14:10 | |
| *** rnirmal has joined #openstack-dev | 14:11 | |
| *** yamahata_lt has joined #openstack-dev | 14:17 | |
| *** salv-orlando has left #openstack-dev | 14:54 | |
| *** BK_man_ has joined #openstack-dev | 15:06 | |
| BK_man_ | hi all. how strict nova deps are? Can I use recent version of eventlet instead of 0.9.12? | 15:07 |
| *** colinnich has joined #openstack-dev | 15:19 | |
| *** BK_man has quit IRC | 15:24 | |
| *** BK_man_ is now known as BK_man | 15:24 | |
| *** BK_man_ has joined #openstack-dev | 15:29 | |
| *** johnpur has joined #openstack-dev | 15:29 | |
| *** ChanServ sets mode: +v johnpur | 15:29 | |
| *** galthaus_ has joined #openstack-dev | 15:36 | |
| *** BK_man_ has quit IRC | 15:45 | |
| *** jmeredit has joined #openstack-dev | 15:50 | |
| *** bcwaldon has quit IRC | 15:56 | |
| openstackjenkins | Project swift build #260: SUCCESS in 37 sec: http://jenkins.openstack.org/job/swift/260/ | 16:01 |
| openstackjenkins | Tarmac: Change the CHANGELOG; now we'll only change it as we cut a release. | 16:01 |
| vishy | BK_man: you will need bleeding edge eventlet | 16:13 |
| vishy | anyone around that can look at my no-db-messaging prototype? | 16:13 |
| vishy | https://code.launchpad.net/~vishvananda/nova/no-db-messaging/+merge/61189 | 16:14 |
| BK_man | vishy: ok. why pip-requires still contains a reference to 0.9.12 ? | 16:15 |
| vishy | not sure :) | 16:18 |
| vishy | you need a patched eventlet though | 16:18 |
| openstackjenkins | Project nova build #905: SUCCESS in 2 min 43 sec: http://jenkins.openstack.org/job/nova/905/ | 16:18 |
| openstackjenkins | Tarmac: Improved error notification in network create | 16:18 |
| vishy | comstud, soren: is the new eventlet patch in the ppa now? | 16:19 |
| comstud | haven't heard that it's in there yet | 16:20 |
| *** scotticus has joined #openstack-dev | 16:20 | |
| BK_man | vishy: I applied eventlet patch and still getting Broken pipe error. | 16:24 |
| BK_man | vishy: http://paste.openstack.org/show/1357/ | 16:25 |
| BK_man | vishy: this is using eventlet-0.9.15+ patch https://bitbucket-assetroot.s3.amazonaws.com/which_linden/eventlet/20110513/87/eventlet-socket-timeout.patch | 16:25 |
| vishy | BK_man: I've never seen that error | 16:26 |
| BK_man | vishy: 3 nodes (cc+compute and two computes), 200 instances | 16:26 |
| vishy | comstud: ^^ Is that the same error you were seeing running 100 instances? | 16:27 |
| vishy | i'm wondering if you need to raise ulimit | 16:27 |
| comstud | checking | 16:27 |
| BK_man | vishy: another error: http://paste.openstack.org/show/1358/ (Socket closed) | 16:27 |
| * BK_man wondering how can I raise it within start-stop-daemon | 16:28 | |
| comstud | no | 16:28 |
| comstud | not seeing Broken Pipe | 16:28 |
| comstud | Broken Pipe implies you have enough file descriptors in ulimit | 16:28 |
| comstud | since the connection was active | 16:29 |
| comstud | i mean, since fd was allocated | 16:29 |
| BK_man | in total: 1 Broken pipe error, 2 Socket closed errors. 3/200 failure rate | 16:29 |
| vishy | i wonder if the broken pipe is a cascading failure from socket close... | 16:30 |
| comstud | i wonder if rabbit is closing the socket immediately on connect | 16:30 |
| comstud | you could see both Socket closed and EPIPE from that | 16:30 |
| comstud | EPIPE if you happen to not catch the socket closed before trying to send() | 16:30 |
| vishy | tr3buchet: ping | 16:33 |
| comstud | that kinda looks to be what's going on | 16:33 |
| vishy | comstud: why would rabbit do such a thing? | 16:33 |
| tr3buchet | vishy: ping | 16:33 |
| comstud | i was seeing it in testing the other day when I was doing something wrong | 16:33 |
| comstud | that something wrong happened to be invalid login credentials | 16:33 |
| vishy | tr3buchet: how are we looking on multi-nic | 16:33 |
| openstackjenkins | Project nova build #906: SUCCESS in 2 min 41 sec: http://jenkins.openstack.org/job/nova/906/ | 16:34 |
| openstackjenkins | Tarmac: Added network_info into refresh_security_group_rules | 16:34 |
| openstackjenkins | That fixs https://bugs.launchpad.net/nova/+bug/773308 | 16:34 |
| vishy | tr3buchet: we only have a couple weeks until milestone 1, I'm hoping to get it in | 16:34 |
| *** jmeredit has quit IRC | 16:34 | |
| comstud | BK_man: did you happen to delete the 'guest' user from rabbitmqserver? | 16:35 |
| tr3buchet | vishy: i've got a few things related to adding networks to project through the api, and then all that's left is making sure the test coverage is there | 16:35 |
| comstud | or set an invalid rabbit_userid/rabbit_password? | 16:35 |
| BK_man | comstud: nope. just installed rpm and started it. not configuration at all | 16:35 |
| tr3buchet | i expect the tests to take good few days at least | 16:35 |
| comstud | BK_man: Hm. rabbitmqctl list_users | 16:36 |
| comstud | see if 'guest' is there | 16:36 |
| comstud | and then see if this matches: | 16:36 |
| comstud | # rabbitmqctl list_user_permissions guest | 16:36 |
| comstud | Listing permissions for user "guest" ... | 16:36 |
| comstud | /.*.*.* | 16:36 |
| comstud | ...done. | 16:36 |
| tr3buchet | vishy: what's the plan for the hyperV, vmware, and ec2 conflicts? | 16:36 |
| comstud | oops, those are supposed to be tabs i guess | 16:36 |
| BK_man | # rabbitmqctl list_users | 16:37 |
| BK_man | Listing users ... | 16:37 |
| BK_man | guest true | 16:37 |
| BK_man | ...done. | 16:37 |
| comstud | ok. | 16:37 |
| comstud | i think i'd next check rabbit logs | 16:37 |
| *** bcwaldon has joined #openstack-dev | 16:37 | |
| BK_man | comstud: the only thing that I do not rip between my test iterations is rabbit config | 16:37 |
| comstud | you can try a quick test program, also | 16:39 |
| comstud | http://paste.openstack.org/show/1362/ | 16:39 |
| vishy | tr3buchet: can you email the list regarding vmware? I think the guys from citrix will help there. I think HyperV we will just have to only support single nic initially. What breaks in 3c2? | 16:40 |
| vishy | s/3c2/ec2 | 16:40 |
| BK_man | comstud: passed silently | 16:41 |
| tr3buchet | vishy: the ec2 api is referring to instance['mac_address'] and instance['fixed_ip'] | 16:41 |
| comstud | BK_man: Hm. Well, that's interesting. | 16:41 |
| vishy | tr3buchet: those should be fairly easy fixes though, no? It can just use the first from the list | 16:42 |
| BK_man | comstud: let me add rabbitmq cleanup code to my cleanup sequenct and repeat test | 16:42 |
| comstud | ok, the only other thing I can suggest right now is looking at rabbit logs, I think | 16:42 |
| comstud | since it really appears it is closing the connection on you | 16:42 |
| tr3buchet | vishy: i can also put shims on the instance model for it. that means ec2 will be single nic also | 16:42 |
| comstud | so it must not like something. | 16:42 |
| tr3buchet | i'm fine with either way | 16:43 |
| tr3buchet | wasn't sure if that was a decision i should make | 16:43 |
| tr3buchet | i put shims on the model for now | 16:43 |
| *** markvoelker has quit IRC | 16:48 | |
| tr3buchet | vishy: there is also the tests for the hyperv vmware and libvirt hypervisors | 16:49 |
| tr3buchet | i'll get an email to you today with some specifics | 16:50 |
| vishy | tr3buchet: thanks | 16:50 |
| openstackjenkins | Project nova build #907: SUCCESS in 2 min 48 sec: http://jenkins.openstack.org/job/nova/907/ | 16:58 |
| openstackjenkins | Tarmac: Convert instance_type_ids in the instances table from strings to integers to enable joins with instance_types. This in particular fixes a problem when using postgresql. | 16:58 |
| *** mszilagyi has joined #openstack-dev | 16:59 | |
| vishy | sandywalsh: ping | 17:01 |
| *** Tv has joined #openstack-dev | 17:02 | |
| *** scotticus has left #openstack-dev | 17:02 | |
| *** dprince has quit IRC | 17:03 | |
| *** Tv has quit IRC | 17:06 | |
| *** dprince has joined #openstack-dev | 17:07 | |
| *** yamahata_lt has quit IRC | 17:07 | |
| *** Tv has joined #openstack-dev | 17:09 | |
| *** jdurgin has joined #openstack-dev | 17:10 | |
| *** mattray has quit IRC | 17:16 | |
| dprince | vishy: just looked at your no-db-messaging branch. Got some errors. | 17:23 |
| *** zaitcev has joined #openstack-dev | 17:23 | |
| dprince | vishy: noted them in the merge prop... | 17:23 |
| vishy | cool thx | 17:23 |
| *** jkoelker has quit IRC | 17:25 | |
| *** jkoelker has joined #openstack-dev | 17:26 | |
| *** dprince_ has joined #openstack-dev | 17:30 | |
| *** dprince_ has quit IRC | 17:30 | |
| *** dprince has quit IRC | 17:33 | |
| *** dprince has joined #openstack-dev | 17:34 | |
| *** blamar has quit IRC | 17:40 | |
| *** jmeredit has joined #openstack-dev | 17:42 | |
| vishy | dprince: interesting error, not quite sure how i broke that | 17:56 |
| termie | comstud: heya, i hear you are interested in connection pools for rabbit, i'm intending to write that up today and wanted to see whether you'd started | 17:57 |
| *** bcwaldon has quit IRC | 17:57 | |
| vishy | dprince: hmm how did that ever work? | 17:57 |
| vishy | dprince: server_management_url != server_manageent_url | 17:57 |
| *** antonyy has joined #openstack-dev | 17:58 | |
| dprince | vishy: looking... | 17:58 |
| comstud | termie: I have started as a part of working on a bug | 17:59 |
| vishy | dprince: not only that but, the . syntax shouldn't be used | 18:00 |
| vishy | it should be token['auth_manageent_url'] | 18:00 |
| comstud | termie: I'm almost finished.. I'll have a branch pushed up soon that I wouldn't mind some feedback on | 18:00 |
| termie | comstud: k, how soon is soon? (it's something blocking me right now) | 18:01 |
| comstud | 2-3 hrs | 18:01 |
| comstud | maybe much less | 18:01 |
| comstud | i have to deal with a couple other distractions real quick first | 18:01 |
| termie | comstud: cool, will wait til then | 18:01 |
| comstud | like my growling tummy | 18:01 |
| *** blamar__ has joined #openstack-dev | 18:01 | |
| comstud | as a part of this, I also have a change to make the 3 topic listeners all use the same connection | 18:02 |
| comstud | with a wait() | 18:02 |
| comstud | instead of these greenthreads that poll every .1 seconds | 18:02 |
| termie | right, that was part of my plan too | 18:02 |
| termie | cool, well sounds like you're on the right track, look forward to codeville | 18:03 |
| comstud | this results in a poll on a single fd.. | 18:03 |
| comstud | instead of these constant send/recvs every .1 seconds for 3 connections | 18:03 |
| comstud | cools :) | 18:03 |
| termie | (unless we're both wrong) | 18:03 |
| comstud | haha, right :) | 18:03 |
| openstackjenkins | Project nova build #908: SUCCESS in 2 min 57 sec: http://jenkins.openstack.org/job/nova/908/ | 18:03 |
| openstackjenkins | Tarmac: Added missing metadata join to instance_get calls. | 18:03 |
| comstud | i'd probably like to re-factor rpc.py a little more than I'm going to for now | 18:04 |
| *** hub_cap has joined #openstack-dev | 18:04 | |
| comstud | just trying minimal disruptions to the current code to see how this works | 18:05 |
| comstud | another thing i'm thinking about... | 18:05 |
| comstud | i wonder if there's any gain to using real threads vs greenlets for some things. | 18:05 |
| comstud | but i've pushed that to the back of my head for now | 18:06 |
| comstud | who's going to this meetup Thurs? | 18:07 |
| comstud | deciding whether or not I want to go | 18:07 |
| comstud | might try to camp at cloudkick during the day if I can find a corner | 18:08 |
| termie | hmm dunno about a meetup, link? | 18:15 |
| termie | we tend to have space at our temp offices | 18:16 |
| sandywalsh | vishy, just got back ... ping back | 18:19 |
| vishy | sandywalsh: just wondering if there is any value to this bp: https://blueprints.launchpad.net/nova/+spec/zones101 | 18:22 |
| vishy | seems like all the work is happening in distributed scheduler, do we need a bp for zone stuff? | 18:22 |
| sandywalsh | vishy, yeah, I'll kill that one and keep it in dist sched. | 18:23 |
| sandywalsh | I can link my summit presentation to the bp if that's sufficient? | 18:23 |
| vishy | i was thinking just marking it superseded by dist sched | 18:24 |
| sandywalsh | sure, that's fine | 18:24 |
| sandywalsh | for my future reference, are these sorts of placeholders required as we move from one release to the next? Or can a previous bp carry over? | 18:25 |
| comstud | termie: http://www.meetup.com/openstack/events/18167311/ | 18:27 |
| *** johan___ has joined #openstack-dev | 18:27 | |
| comstud | you guys going to be at your temp space on Thursday? | 18:27 |
| johan___ | merge for LP778678 failed because of jenkins failures unrelated to the patch. does anyone know how to fix jenkins? | 18:28 |
| pyhole | LP Bug #778678: nova.virt.xenapi.vmops _run_ssl() should write directly to stdin instead of file [Status: In Progress, Assignee: Johannes Erdfelt] https://bugs.launchpad.net/bugs/778678 | 18:28 |
| termie | comstud: 7pm, yikes | 18:29 |
| *** blamar has joined #openstack-dev | 18:29 | |
| comstud | yeah, it's kinda late | 18:29 |
| dprince | vishy: I'm filing a ticket now. | 18:29 |
| termie | comstud: yeah, i expect we'll be here | 18:29 |
| dprince | vishy: the column name is incorrect in auth_tokens. | 18:29 |
| dprince | vishy: in your code you changed the NovaBase model to reflect on the database table instead of using the model definition in models.py. | 18:30 |
| comstud | might be worthwhile to come out there, if there's anything you want to go over with rabbit connection pooling, anyway | 18:30 |
| dprince | vishy: SInce the column is just plain wrong in the DB your branch breaks auth for the OSAPI. Pretty sweet hey! | 18:30 |
| termie | comstud: worth it anyway, i'd say, good to co-hackl | 18:30 |
| comstud | yeah | 18:31 |
| markwash | dev question: what would you think if I changed quotas so that if the deployer specifies --quota_instances=-1, instances are unlimited by default | 18:31 |
| comstud | i'll ping you guys for the address | 18:31 |
| comstud | close to bart? | 18:31 |
| comstud | can't remember where vishy told me it was | 18:31 |
| comstud | or maybe it was jesse | 18:32 |
| blamar | markwash, also the lack of the flag would default to -1 aka unlimited? | 18:32 |
| termie | comstud: yeah, close to bart, fremont and howard, get off at embarcadero, take the away from the bay exit and walk down fremont | 18:32 |
| vishy | dprince: https://code.launchpad.net/~vishvananda/nova/fix-token | 18:32 |
| vishy | dprince: not sure if it is worth a db migration or not? | 18:33 |
| termie | comstud: address == search for rocketspace | 18:33 |
| markwash | blamar: the lack of a flag would probably still mean 10 instances | 18:33 |
| vishy | comstud: even closer to the bus station | 18:33 |
| markwash | since that is the default value of that flag | 18:33 |
| dprince | vishy: https://bugs.launchpad.net/nova/+bug/784207 | 18:33 |
| comstud | some amount if walking is probably good for me | 18:33 |
| comstud | :) | 18:33 |
| comstud | of | 18:33 |
| westmaas | termie: really? that is so so awesome. we had a guy apply for a job here and call us rocketspace. | 18:33 |
| vishy | dprince: especially since that code will go away when we switch to keystone | 18:34 |
| termie | westmaas: hmm? | 18:34 |
| dabo | westmaas: maybe they heard we were working with NASA and got confused? | 18:34 |
| termie | westmaas: called rackspace rocketspace? | 18:35 |
| westmaas | termie: yes | 18:35 |
| vishy | comstud: 181 fremont | 18:35 |
| comstud | yep, got it. | 18:35 |
| comstud | need anything special to get in? | 18:35 |
| comstud | secret handshake? | 18:36 |
| termie | comstud: circumcision | 18:36 |
| dprince | vishy: Even so I'd lean towards a migration. | 18:36 |
| termie | doesn't hurt as much as you'd expect | 18:36 |
| vishy | dprince: ok I started writing one | 18:36 |
| comstud | haha | 18:37 |
| blamar | limits don't make any sense to be flags, they're deployment options which should be in the database and not defined in two places (flags + database) | 18:37 |
| dprince | vishy: Cool. Wanna take that ticket then too? | 18:37 |
| comstud | i prob won't make it out til after 11am | 18:38 |
| markwash | blamar: I'm not necessarily disagreeing, just kind of trying to make a smaller change at this time--i.e. I would keep the quota config flags | 18:38 |
| dprince | vishy: Your branch looks Okay otherwise however I'd like to smoke test it again when we get this migration in. | 18:38 |
| vishy | ok | 18:38 |
| vishy | I will work on migration | 18:38 |
| vishy | just have to remember how to query db in sqlalchemy :) | 18:39 |
| comstud | or later, i guess. i have a mtg until 10:30 | 18:39 |
| comstud | and it's probably an 1h:15m trip for me | 18:40 |
| comstud | can be there in time for lunch at the gold club. | 18:40 |
| termie | hoyoo | 18:41 |
| comstud | :) | 18:41 |
| dprince | markwash: -1 seems Okay for the default flag. I'd look for prior art in nova for an 'unlimited' default flag. | 18:42 |
| comstud | ok, gonna make some lunch and then get some code pushed up | 18:43 |
| termie | DO IT | 18:43 |
| termie | looking forward to it, i mean | 18:43 |
| comstud | been prototyping outside of nova, so I need to merge it in | 18:43 |
| comstud | mostly cut n paste, though | 18:44 |
| dprince | markwash: Null for unlimited also seems fine. | 18:44 |
| comstud | ok bbs | 18:44 |
| * comstud & | 18:44 | |
| markwash | dprince: yeah actually I like that better, but there doesn't seem to be a way to get DEFINE_integer to accept the string None | 18:44 |
| termie | markwash: many people use 0 or -1 for unlimited limits | 18:54 |
| markwash | termie: cool, that is also the simplest way for me to take care of this, so I'm going for it | 18:55 |
| *** bcwaldon has joined #openstack-dev | 19:00 | |
| vishy | any sqlalchemy geniuses know how to reference another column in an update statement? | 19:03 |
| ttx | Meeting in ~2 hours in #openstack-meeting | 19:03 |
| dprince | vishy: Would something like we did in migration 014 work? | 19:06 |
| markwash | I think the value in the update can be a select statement | 19:09 |
| markwash | err nm | 19:09 |
| markwash | false | 19:09 |
| termie | these conversations are so terrifying to me after having managed the database for a high traffic production site | 19:10 |
| termie | every one sounds like "24 hours of downtime" | 19:10 |
| markwash | yeah I get that feeling too | 19:10 |
| vishy | got it | 19:11 |
| *** rnirmal has quit IRC | 19:38 | |
| openstackjenkins | Project nova build #909: SUCCESS in 2 min 44 sec: http://jenkins.openstack.org/job/nova/909/ | 19:53 |
| openstackjenkins | Tarmac: Fixes improper attribute naming around instance types that broke Resizes. | 19:53 |
| ttx | Meeting in one hour in #openstack-meeting ! | 19:59 |
| *** mattray has joined #openstack-dev | 19:59 | |
| *** dprince has quit IRC | 20:05 | |
| openstackjenkins | Project nova build #910: SUCCESS in 2 min 47 sec: http://jenkins.openstack.org/job/nova/910/ | 20:13 |
| openstackjenkins | Tarmac: Docstring cleanup and formatting (nova/network dir). Minor style fixes as well. | 20:13 |
| soren | vishy: Not yet. Will be this evening at some point. There was a surprising amount of work involved in updating to 0.9.15 in Ubuntu. | 20:15 |
| *** galthaus_ has quit IRC | 20:16 | |
| soren | vishy, comstud: Any particular version of Ubuntu you're most interested in? | 20:20 |
| * soren has just had an awesome idea | 20:21 | |
| *** salv-orlando has joined #openstack-dev | 20:23 | |
| nhm | soren: does it involve alcohol? | 20:26 |
| soren | nhm: It should! | 20:27 |
| *** alex-meade has quit IRC | 20:30 | |
| nhm | soren: great, I vote to improve the already good idea with liquor. | 20:30 |
| bcwaldon | seconded | 20:30 |
| *** antonyy has quit IRC | 20:36 | |
| vishy | dprince: https://code.launchpad.net/~vishvananda/nova/fix-token/+merge/61304 | 20:37 |
| comstud | termie: lp:~cbehrens/nova/rpc-connection-pool | 20:53 |
| termie | comstud: sweeet, checking it out | 20:53 |
| comstud | tell me how nasty this is | 20:54 |
| termie | so nasty | 20:54 |
| termie | you bad boy | 20:54 |
| comstud | :) | 20:54 |
| * termie got weird | 20:54 | |
| comstud | i've not run tests yet | 20:54 |
| termie | that's not promising | 20:54 |
| comstud | other than i tested my connectionpool class standalone | 20:54 |
| comstud | yeah, well, it's a wip | 20:54 |
| comstud | figured i'd get you something sooner | 20:54 |
| comstud | :) | 20:54 |
| termie | cool, thanks :) | 20:54 |
| comstud | np | 20:55 |
| comstud | i can be talked out of using a pool of threads for this, but i kinda like the idea | 20:56 |
| termie | comstud: i don't see what is to be gained from threads... | 20:56 |
| comstud | there's some work needed around handling reconnection to rabbit on errors still | 20:56 |
| comstud | there's probably no gain | 20:57 |
| comstud | although what this gives you is.. | 20:58 |
| comstud | an easy way to close idle connections when not in use | 20:58 |
| ttx | Meeting starts in 2 min. in #openstack-meeting | 20:58 |
| comstud | but i suppose some sort of timer could do that also | 20:58 |
| comstud | there's some tunables that should be flags that are just hardcoded right now, also | 20:59 |
| comstud | if this sucks, we can re-do it entirely. i won't be offended. | 21:00 |
| termie | no worries, about to take a look at it | 21:00 |
| comstud | i think the 'ConsumerSet' for the topic listeners is what we want, tho | 21:00 |
| termie | comstud: i was also looking at kombu | 21:01 |
| termie | comstud: 'cause i guess that is the new version of carrot | 21:01 |
| comstud | ahhh | 21:01 |
| comstud | i'm kinda annoyed with carrot. | 21:01 |
| comstud | i'll have to take a peak at that sometime | 21:02 |
| termie | it does some things in a way that would be much more wonderful if it was designed to be used with eventlet to begin with | 21:03 |
| comstud | ah | 21:03 |
| comstud | import eventlet.queue missing | 21:06 |
| comstud | bah, number of fixes to make | 21:06 |
| *** uvirtbot has joined #openstack-dev | 21:06 | |
| *** bcwaldon has quit IRC | 21:06 | |
| soren | bug 123456 | 21:06 |
| uvirtbot | Launchpad bug 123456 in xine-lib "podcast crashes amarok" [Undecided,Fix released] https://launchpad.net/bugs/123456 | 21:07 |
| soren | \o/ | 21:07 |
| *** bcwaldon has joined #openstack-dev | 21:07 | |
| jk0 | lp123456 | 21:07 |
| pyhole | LP Bug #123456: podcast crashes amarok [Status: Fix Released, Assignee: None] https://bugs.launchpad.net/bugs/123456 | 21:07 |
| uvirtbot | Launchpad bug 123456 in xine-lib "podcast crashes amarok" [Undecided,Fix released] | 21:07 |
| jk0 | pyhole > uvirtbot | 21:07 |
| comstud | pyhole doesn't tell the project name | 21:08 |
| comstud | :) | 21:08 |
| comstud | fix it! | 21:08 |
| jk0 | good call | 21:08 |
| *** markvoelker has joined #openstack-dev | 21:12 | |
| dabo | comstud: check out celery. What I saw of it at PyCon looked much better than carrot. | 21:12 |
| comstud | yeah, i've looked at it somewhat | 21:13 |
| comstud | ok, after fixing typos and junk, i have a real failure now | 21:14 |
| comstud | ah | 21:16 |
| termie | dabo: celery is sort of a bigger change | 21:16 |
| termie | dabo: we've talked about it a lot | 21:16 |
| termie | dabo: not ruled out but requires a pretty fundamental re-organization of everything | 21:16 |
| dabo | I know. The biggest argument in favor of carrot is that it would be a big change to go to anything else. | 21:17 |
| termie | dabo: except for kombu :p | 21:17 |
| dabo | haven't looked at kombu | 21:17 |
| termie | dabo: it is carrot compatible, by same guy just newer code | 21:17 |
| dabo | just figured that as long as comstud was messing around in the internals, he should look at all options. | 21:17 |
| dabo | termie: you mean rather than updating carrot? | 21:18 |
| termie | dabo: carrot was discontinued for whatever reason and kombu replaced it, yes | 21:19 |
| comstud | yeah, a lot of it looks the same | 21:20 |
| dabo | that sucks. Even more reason to move to kombu, then | 21:20 |
| comstud | i can't actually spot a difference at first glance | 21:20 |
| termie | anyway, re celery, it is getting closer to what we want both because it has improved and because the project is moving closer to wanting a task queue | 21:20 |
| termie | but it will require a much bigger push than what we're working on today | 21:20 |
| *** pyhole has quit IRC | 21:22 | |
| *** pyhole has joined #openstack-dev | 21:22 | |
| jk0 | lp123456 | 21:22 |
| pyhole | LP Bug #123456 in xine-lib (Ubuntu): "podcast crashes amarok" [Status: Fix Released, Assignee: None] https://bugs.launchpad.net/bugs/123456 | 21:22 |
| uvirtbot | Launchpad bug 123456 in xine-lib "podcast crashes amarok" [Undecided,Fix released] https://launchpad.net/bugs/123456 | 21:22 |
| uvirtbot | Launchpad bug 123456 in xine-lib "podcast crashes amarok" [Undecided,Fix released] | 21:22 |
| jk0 | done | 21:22 |
| comstud | :) | 21:23 |
| termie | comstud: okay, looking at your stuff now, my original plan had just been to use an eventlet.pools.Pool subclass and make it provide Connection instances | 21:25 |
| comstud | i was looking at that, also | 21:25 |
| termie | comstud: using the existing interface | 21:25 |
| comstud | trying to remember why I didn't go that way | 21:25 |
| termie | so that .instance() is effectively a .get() | 21:25 |
| comstud | some of the complications are from trying to handle errors and reconnecting | 21:26 |
| termie | comstud: looks to me like the existing mechanisms can all be made to work the same way | 21:27 |
| comstud | hmmm yeah. | 21:29 |
| termie | comstud: i'll hack on it a bit and see if i can provide an example of what i am thinking | 21:31 |
| comstud | i think the hard thing about that was trying to timeout old connections | 21:31 |
| comstud | say | 21:31 |
| comstud | you define a max # of conns you want | 21:31 |
| comstud | if you reach the max.. during bursting.. | 21:31 |
| comstud | it'd be nice when things settled down that you could close some | 21:31 |
| comstud | not sure how easy that might be to implement with pools.Pool | 21:31 |
| *** hub_cap has quit IRC | 21:32 | |
| comstud | i was also just considering keeping an array of connections | 21:32 |
| comstud | and always pop, append | 21:33 |
| comstud | timer can step through and remove old ones not used since 'x' | 21:33 |
| comstud | i blame overcomplicating this on it having been 1am and i misread something. | 21:41 |
| * comstud smirks. | 21:41 | |
| * termie shrugs | 21:41 | |
| termie | i had some pretty specific ideas in mind, too, so i'm mostly trying to match those | 21:42 |
| comstud | well, it actuall turns out i think this does what i wanted, also, but i didn't thin it did | 21:43 |
| openstackjenkins | Project nova build #911: SUCCESS in 2 min 42 sec: http://jenkins.openstack.org/job/nova/911/ | 21:43 |
| openstackjenkins | Tarmac: Fixes the naming of the server_management_url in auth and tests. | 21:43 |
| comstud | eventlet (and python) is new to me.. i'm really a C person:) | 21:44 |
| * Vek affirms--comstud *is* a C person | 21:45 | |
| comstud | :) | 21:45 |
| * jk0 prefers PHP | 21:45 | |
| jk0 | :D | 21:45 |
| comstud | riiight | 21:45 |
| jk0 | jk :P | 21:45 |
| Vek | I've programmed in PHP, Perl, Python, Bourne shell, C, JavaScript, even a little Java. | 21:46 |
| Vek | (and yes, I mean I *programmed* in Bourne shell. A recursive dependency resolver, no less. *not* fun...but boy I wish I'd saved that code! ;) | 21:46 |
| comstud | i think i have pools.Pool working.. rpc tests pass | 21:57 |
| comstud | all tests pass | 21:58 |
| comstud | i think there's 1 problem here tho | 21:59 |
| comstud | the spawn in AdapterConsumer.receive() | 21:59 |
| comstud | can't really spawn until after the message has been acked | 22:00 |
| *** alekibango has joined #openstack-dev | 22:01 | |
| *** mgius has joined #openstack-dev | 22:02 | |
| *** johnpur has quit IRC | 22:04 | |
| vishy | tv: ryu is working on a milder refactor of the exiting network code. | 22:05 |
| vishy | tv: if we can be clear the specific calls that melange/dhcp service needs to support he should be able to put the proper stuff in | 22:05 |
| *** letterj has joined #openstack-dev | 22:06 | |
| *** ChanServ sets mode: +v letterj | 22:06 | |
| Tv | vishy: can you skim this quick: https://github.com/tv42/melange-discovery/blob/master/melange-discovery.rst | 22:06 |
| *** hub_cap has joined #openstack-dev | 22:06 | |
| vishy | tv: looks interesting | 22:07 |
| Tv | vishy: so the basic idea is to push the decision making into the IPAM service | 22:08 |
| vishy | tv: currently working on a "cloud config drive" which could replace the need for virtio 9p fs | 22:08 |
| Tv | vishy: oh i'm actually interested in details of that | 22:08 |
| Tv | vishy: because i think the virtiofs was pretty neat ;) | 22:08 |
| vishy | tv: essentially it attaches a fat16 drive that contains config info | 22:08 |
| Tv | vishy: boot time only, then? | 22:08 |
| vishy | tv: undetermined. Looking at the possiblity of making it read/write by host and guest | 22:09 |
| vishy | initially for boot time config though | 22:09 |
| Tv | vishy: that's pretty much impossible, as evidenced by usb-connected smartphones etc | 22:09 |
| vishy | tv: interesting | 22:10 |
| vishy | tv: another possiblity is just to use a host only nic | 22:10 |
| Tv | vishy: i don't really like the idea of needing networking to configure networking ;) | 22:10 |
| comstud | host and guest can't mount fat16 at same time | 22:10 |
| vishy | :( | 22:11 |
| comstud | you'd need a distributed fs | 22:11 |
| comstud | but | 22:11 |
| comstud | you could use the block device | 22:11 |
| soren | I think we really, *really* need to come up with a stack of use cases. | 22:11 |
| comstud | raw block device to do something | 22:11 |
| Tv | but why not use something like virtfs, then | 22:11 |
| *** blamar__ has quit IRC | 22:11 | |
| soren | ...and determine which are valid and which aren't, and then see how block device, network interface, serial interface, etc. fit those use cases. | 22:12 |
| Tv | i mean, i understand it's virtio-specific for now, but 9p can be transported over any stream.. | 22:12 |
| comstud | soren, sure | 22:12 |
| soren | So far, I feel the discussion has been very vague ("what if I want to send sometihng back?" What is "something", really?) | 22:12 |
| soren | Apart from network config. | 22:12 |
| comstud | i'm not proposing any particular solution yet | 22:12 |
| soren | ...which has been quite clear. | 22:12 |
| comstud | :) | 22:12 |
| comstud | yea | 22:12 |
| vishy | tv: configuring guests should use something that is extremely compatible | 22:12 |
| vishy | tv: i think a special nic device that dhcps from host could work | 22:13 |
| Tv | vishy: sort of yes, but on the extreme side of that is aws-level "compatible with everything" that is hard to use for many people | 22:13 |
| Tv | e.g. you can't transport secrets nicely with aws user-data | 22:13 |
| soren | I think that particular use case is weird, though. In general, configuring network cards dynamically *so* a ssolved problem. | 22:14 |
| soren | Tv: Only because there's no way to turn it off. | 22:14 |
| Tv | soren: yeah network cards is very easy, think about things like SSL keys and certs instead | 22:14 |
| soren | Tv: If there was a way to just tell the meta-data server that you're done with it, it would be ok. You could grab it at boot time, store it on disk readable only by root, disable the service, done. | 22:15 |
| Tv | soren: yeah.. the other way around is "hey customer management node, i just created an ssh host key for myself, here it is" -- how do you build trust for that | 22:16 |
| soren | Tv: GetConsoleOutput? | 22:17 |
| Tv | parse the potentially big stream of text for a magic cookie and a base64-encoded blob? | 22:18 |
| soren | Yes. | 22:18 |
| Tv | ugly but doable, doable but ugly ;) | 22:18 |
| soren | It's not super elegant, but it's delightfully extensible. | 22:19 |
| *** jkoelker has quit IRC | 22:26 | |
| *** jkoelker has joined #openstack-dev | 22:27 | |
| openstackjenkins | Project swift build #261: SUCCESS in 29 sec: http://jenkins.openstack.org/job/swift/261/ | 22:31 |
| openstackjenkins | Tarmac: Use bracketed notation for ip literals in rsync calls, so compressed ipv6 literals work. | 22:31 |
| *** mattray has quit IRC | 22:32 | |
| *** jmeredit has left #openstack-dev | 22:33 | |
| *** mszilagyi has quit IRC | 22:51 | |
| comstud | termie: lp:~cbehrens/nova/rpc-improvements | 22:55 |
| termie | comstud: hmm, seems to back out most of your other changes? | 22:59 |
| comstud | yeah | 23:00 |
| comstud | this probably makes more sense | 23:00 |
| comstud | easier | 23:00 |
| comstud | and I can implement the timeouts i want around this later | 23:00 |
| comstud | hrm | 23:01 |
| comstud | nova/service.py:95:9: E303 too many blank lines (2) | 23:01 |
| termie | comstud: doesn't look like you are using the pool anywhere | 23:01 |
| comstud | ConnectionPool.get ? | 23:01 |
| comstud | look @ 'def call' | 23:01 |
| termie | ah | 23:02 |
| termie | but not in service? | 23:02 |
| comstud | i didn't make Service use it | 23:02 |
| termie | i was thinking actually swapping out the implementations | 23:02 |
| comstud | it's only 1 connection now | 23:02 |
| comstud | (Service) | 23:02 |
| comstud | vs 3 | 23:02 |
| comstud | everything else will be used out of the pool | 23:02 |
| comstud | the tests seem to be failing, though, due to fakerabbit Backend not supporting ConsumerSet properly. | 23:03 |
| comstud | (multiplexing 3 consumers onto 1 connection while listening for events for all 3) | 23:03 |
| termie | you also have a single line that says "fuck" | 23:03 |
| comstud | LOL | 23:03 |
| comstud | probably. | 23:03 |
| comstud | where is that? | 23:04 |
| termie | in consumerset.wait | 23:04 |
| comstud | strange | 23:04 |
| * comstud finds that kinda funny. | 23:04 | |
| comstud | pushed some comment updates, pep8, and fuck removal. | 23:06 |
| comstud | looks like i missed fanout_cast() | 23:06 |
| comstud | can build instance despite test failing. | 23:10 |
| comstud | anyway | 23:10 |
| comstud | This more like what you were looking for? | 23:10 |
| termie | aye | 23:10 |
| termie | not sure what is required to make fakerabbit work | 23:10 |
| comstud | I'm cool with it working like this | 23:10 |
| comstud | yeah | 23:10 |
| comstud | i only briefly looked at it so far | 23:11 |
| comstud | but it looks like fakerabbit.Backend.consume() only looks at a single queue | 23:11 |
| comstud | even tho multiple are registered | 23:11 |
| comstud | probably not a difficult fix | 23:11 |
| comstud | yeah | 23:12 |
| comstud | ok. | 23:12 |
| termie | do you want to look at fixing fake rabbit or should i? | 23:13 |
| comstud | i see what needs to be done | 23:13 |
| comstud | at least initially | 23:14 |
| termie | there are a few tests using mocks that are going to be confused as well | 23:14 |
| comstud | taking a stab at it real quick | 23:14 |
| comstud | let's see what this does | 23:20 |
| termie | . comstud has disconnected | 23:21 |
| comstud | fail | 23:23 |
| *** bcwaldon has quit IRC | 23:27 | |
| *** hub_cap_ has joined #openstack-dev | 23:28 | |
| *** hub_cap_ has quit IRC | 23:28 | |
| *** hub_cap has quit IRC | 23:31 | |
| comstud | argh | 23:33 |
| comstud | was sitting here testing in an old branch | 23:33 |
| *** Tushar has joined #openstack-dev | 23:34 | |
| termie | i hate that | 23:35 |
| *** mgius has quit IRC | 23:37 | |
| comstud | ok | 23:40 |
| comstud | looks like i got that test to pass | 23:40 |
| comstud | let's see if i broke anything else | 23:40 |
| *** markvoelker has quit IRC | 23:50 | |
| comstud | one last time | 23:51 |
| comstud | ///looks like there was a bug in fakerabbit anyway, even without any changes.. just wasn't exerised | 23:52 |
| comstud | exercised | 23:52 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!