Monday, 2011-06-20

*** bonzay has joined #openstack-dev00:22
*** Binbin has joined #openstack-dev00:47
*** HugoKuo_ has joined #openstack-dev00:59
*** HugoKuo has quit IRC01:03
*** robinbb has joined #openstack-dev01:07
*** Binbin has quit IRC01:08
*** winston-d has joined #openstack-dev01:09
*** bonzay has quit IRC01:30
*** cloudgroups has joined #openstack-dev01:36
*** cloudgroups has left #openstack-dev01:39
openstackjenkinsProject nova build #1,024: SUCCESS in 2 min 57 sec: http://jenkins.openstack.org/job/nova/1024/01:54
openstackjenkinsTarmac: Check that server exists when interacting with /v1.1/servers/<id>/meta resource01:54
*** elasticdog has quit IRC05:11
*** elasticdog has joined #openstack-dev05:13
*** Binbin has joined #openstack-dev06:03
*** reidrac has joined #openstack-dev07:15
*** Binbin has quit IRC07:48
*** Binbin has joined #openstack-dev09:07
*** adiantum_ has joined #openstack-dev09:30
*** adiantum_ has quit IRC09:37
*** Binbin has quit IRC09:41
*** adiantum_ has joined #openstack-dev09:41
*** Daviey has quit IRC09:47
*** adiantum_ has quit IRC09:53
*** Daviey has joined #openstack-dev09:55
*** adiantum_ has joined #openstack-dev09:56
*** adiantum_ has quit IRC10:08
*** adiantum_ has joined #openstack-dev10:14
*** adiantum_ has quit IRC10:24
*** adiantum_ has joined #openstack-dev10:36
*** markvoelker has joined #openstack-dev10:48
*** adiantum_ has quit IRC11:05
*** adiantum_ has joined #openstack-dev11:18
*** adiantum_ has quit IRC11:44
*** adiantum_ has joined #openstack-dev11:45
*** adiantum_ has quit IRC11:56
*** gregm has joined #openstack-dev12:01
*** adiantum_ has joined #openstack-dev12:02
*** adiantum_ has quit IRC12:10
ttxnotmyname: let me know if swift 1.4.1 is OK for release, and I'll process it when I have a break in the conference12:31
ttxnotmyname: preparing a final version branch for your approval, if the wifi here allows12:36
ttxnotmyname: if you're OK for release, then review and accept: https://code.launchpad.net/~ttx/swift/final-1.4.1/+merge/6518812:43
*** gregm has quit IRC12:47
*** gregm has joined #openstack-dev12:56
notmynamettx: ack13:05
*** ameade has joined #openstack-dev13:17
*** bcwaldon has joined #openstack-dev13:22
*** adiantum_ has joined #openstack-dev13:41
*** zul has joined #openstack-dev13:41
*** bcwaldon has quit IRC13:44
*** binbin has joined #openstack-dev13:51
*** jkoelker has joined #openstack-dev13:54
jaypipesttx: ping13:55
ttxjaypipes: pong14:21
*** bcwaldon has joined #openstack-dev14:24
notmynamettx: just got the ok from our QA. everything looks good for 1.4.1. Marking the merge now14:35
ttxnotmyname: cool. I'll process it soon14:36
*** johnpur has joined #openstack-dev14:47
*** ChanServ sets mode: +v johnpur14:47
*** bcwaldon has quit IRC14:54
*** bcwaldon has joined #openstack-dev14:56
*** dragondm has joined #openstack-dev15:03
*** robinbb has left #openstack-dev15:11
*** reidrac has quit IRC15:16
ttxvishy: ping15:25
*** cp16net_ has joined #openstack-dev15:25
vishyttx: pong15:26
vishyttx: back from japan/vacation!15:27
ttxvishy: good!15:27
*** cp16net_ has quit IRC15:27
*** heckj has joined #openstack-dev15:37
*** jaypipes has quit IRC15:40
*** binbin has quit IRC15:44
*** bcwaldon has quit IRC15:51
*** bcwaldon has joined #openstack-dev15:53
*** jaypipes has joined #openstack-dev15:53
*** rnirmal has joined #openstack-dev15:57
vishytr3buchet: ping16:03
tr3buchetvishy: hiya vish16:04
tr3buchetvishy etc16:04
vishytr3buchet: hey, just noticed that you have an except IntegrityError in the multi_nic branch which is undefined16:05
vishyin _allocate_mac_address16:05
tr3buchetvishy: have a paste?16:06
vishyline 45416:06
vishyin network/manager.py (at least that is the line in the last code you pushed)16:06
tr3buchetlooking16:07
tr3buchetvishy: nice catch fixed. pushed.16:11
*** mattray has joined #openstack-dev16:11
vishytr3buchet: I'm planning a new networking mode as well.  Just trying to wrap my head around your stuff so it doesn't conflict.16:13
vishytr3buchet: you have each network belonging to one network host, yes?16:13
vishytr3buchet: I'm planning a mode where every compute host will have nova-network running, and they will all share the same network...16:14
tr3buchetvishy: yes each network belongs to a host, once a host picks it up and configures itself16:16
vishytr3buchet: and it configures itself as gateway/dhcp server for that network?16:16
tr3buchetvishy: yes16:16
tr3buchetonce that's set, that network then becomes usable to nova16:17
tr3buchetin otherwords unconfigured networks are ignored16:17
vishytr3buchet: cool, I'm going to do an HA flat dhcp mode where every compute host gets an ip in the network and acts as gateway/dhcp server for its own vms16:17
vishyi suppose it should be a configuration option on the network itself...16:18
vishyrather than selecting a single host, it will be shared.  And it will need a number of reserved ips for the hosts...16:19
openstackjenkinsProject nova build #1,025: SUCCESS in 2 min 58 sec: http://jenkins.openstack.org/job/nova/1025/16:24
openstackjenkinsTarmac: This patch adds support for working with instances by UUID in addition to integer IDs.16:24
openstackjenkinsThe Zone Scheduler routing mechanics were changed slightly so that when an UUID is passed in, it checks to see whether the item is available locally.16:24
openstackjenkinsIf it isn't it re-routes to a child zone.16:24
openstackjenkinsIf it is available locally, it substitutes the UUID with the integer ID and calls the wrapped function. This is the 'trick' allows us to not change any of the virt-layer code-- everything still uses integer IDs locally.16:24
tr3buchetvishy: interesting16:31
tr3buchetvishy: making a new manager?16:32
vishytr3buchet: originally it was going to be a manager16:34
vishynow it seems like I might have to change the network itself16:34
vishytr3buchet: how does one specify which networks the vm is supposed to connect to?16:34
heckjvishy: I think I found a small comment error - made a change, wanted to check to see if it was correct: https://code.launchpad.net/~heckj/nova/commentfix/+merge/6523016:36
vishytr3buchet: your IntegrityError change won't work16:36
vishyIntegrityError won't get out of the db layer without being wrapped, + it shouldn't be exposed to the network manager anyway...16:37
ttxnotmyname: ok, Swift 1.4.1 release is complete. You can send the announce.16:37
vishyheckj: your fix is correct16:37
heckjK - thanks.16:37
tr3buchetvishy: ok i'll have it raise something else in the sqlalchemy.api16:38
vishytr3buchet: yes probably wrap it with another error16:38
vishytr3buchet: did you see my other question?  How do I specify number of nics and attached networks for instances?16:39
*** gregm has quit IRC16:39
tr3buchetvishy: i updated the comment16:40
tr3buchetvishy: the networks are determined by a function _get_networks_for_instance()16:41
tr3buchetcurrently they don't really take any arguments, I made these to be expanded upon16:41
tr3buchetthere should be one in the NetworkManager superclass and one in the VlanManager subclass16:41
vishytr3buchet: ah ok, so it currently just attaches every instance to all networks16:42
tr3buchetif flat, correct16:42
tr3buchetif there are 5 configured networks you get 5 vifs and an ip from each16:42
vishytr3buchet: in vlan mode, do you have to associate projects and networks in advance?16:42
tr3buchetvishy: you do not16:43
tr3bucheti've made a function to force add additional networks, but it isn't yet exposed to the api16:43
vishyok i see16:44
vishylooks like we will need a network/instance association eventually16:45
tr3buchetyeah it needs to be improved16:46
tr3buchetbut that may be good enough for nova sans NaaS16:46
*** mgius has joined #openstack-dev16:47
vishysigh security groups are a mess too... :(16:47
vishytr3buchet: I'm wondering about making network_id nullable in VirtualInterface16:49
vishytr3buchet: it looks like your relationships need primary_joins as well or they won't handle being deleted very well16:51
tr3buchetvishy: i thought about making it nullable, why would you want it to be nullable?16:57
tr3buchetvishy: anything you'd want to nullify it for i considered solving by making a network for16:57
vishybuilds are failing16:57
tr3buchetvishy: comment updated, and virtual interface exception handling corrected16:58
vishytr3buchet: I suppose i'm thinking about having nics that aren't connected to a network16:58
tr3buchetvishy: right, so in that case, why not create a row in the network table for those nics16:59
*** jdurgin has joined #openstack-dev17:00
tr3buchetvishy: i'm a bit fuzzy on the nics with no network connectivity though. what's the use case?17:00
vishytr3buchet: that seems a little counterintuitive17:00
tr3buchetvishy: i see your point17:00
vishytr3buchet: so you can connect nics to different networks at run time17:00
tr3bucheti guess I'm thinking about it from xen point of view, can you create a vif with no bridge?17:01
vishytr3buchet: it might be out of scope for current networking mode, but I think it might be closer to what NAS is planning17:01
vishytr3buchet: sure you would just have a tap device not bridged into anything17:02
vishylooks like the builders need a new version of sqlite or some such?17:03
tr3buchetvishy: in that scenario the tap device in on the vm correct?17:07
vishytr3buchet: tap device on the host17:08
tr3buchetvishy: and the instances get a vif connected to the tap device17:09
vishyso we need a new carrot to fix the build17:09
vishysoren: ping17:09
vishymtaylor: ping17:09
heckjtr3buchet: thanks for pulling in the update17:14
tr3buchetheckj: no problem, thanks for finding it17:14
heckjreading and trying to get up to speed with the networking17:15
mtaylorvishy: pong17:20
*** rnirmal has quit IRC17:21
vishymtaylor: hey, the builds are all failing due to a failing test.  I think we need to build with carrot 10.5 instead of 10.317:23
mtaylorvishy: ok. the tarmac builds? lemme look real quick17:25
vishyyeah the ubuntu packaging17:25
mtaylork. I'm on it17:25
vishycool thx17:25
comstudyes17:28
comstud10.5 works with it17:29
comstudone of my guys was asking about it last week17:29
comstudwe foudn 10.3 failed, 10.5 works17:29
jk0I think the test that is failing now is unrelated17:30
jk0getting AssertionError: u'\\/dev\\/vdc' != '/dev/vdc' in test_cloud17:30
jk0nm17:32
mtaylorvishy: carrot 10.7 is latest released... still go with 10.5?17:33
vishylatest is fine17:33
mtaylork. cool.17:34
jk0it's still failing on 10.7 for me17:35
vishyo17:36
mtaylorjk0: does 10.5 work and 10.7 not? or just your failure is a non-releated error?17:37
jk0haven't tried 10.5 yet but others report it working17:37
mtaylorhrm17:37
mtaylorthat's vexing17:37
mtaylorok. I've upgraded to 10.5 on the jenkins box17:48
*** gregm has joined #openstack-dev17:48
*** rnirmal has joined #openstack-dev17:55
vishythe really vexing thing is that it passed fine when merging17:56
vishyit is the launchpad builders that are failing17:56
vishyfind . -name disk* | xargs -n1 qemu-img info | grep backing17:56
vishyer whoops17:56
vishyhttps://launchpadlibrarian.net/73828348/buildlog_ubuntu-natty-i386.nova_2011.3~d2~20110620.1200-0ubuntu0ppa1~natty1_FAILEDTOBUILD.txt.gz17:56
vishyand it is only the i386 builds that failed17:57
*** adiantum_ has quit IRC18:09
mtaylorAH - that is vexing ...18:09
mtaylorwell - those are natty builders - and natty has 10.7 of python-carrot18:10
*** gregm has quit IRC18:12
vishyall 3 i386 builds failed18:13
*** johnpur has quit IRC18:13
vishymtaylor: https://blueprints.launchpad.net/nova/+spec/testing-jenkins-integration18:20
vishythink this will be done by diablo-2? or should i push?18:21
mtaylorvishy: lemme re-write that - some of it will be done by diablo-2, but other bits really need their own blueprint. I'll get that done this morning for you18:22
vishymtaylor: excellent thanks18:22
vishymarkwash: do you think we should move the wikis about testing into the docs? or just link to them?18:26
mtaylorvishy: ok - I put a 10.5 backport into our ppa and re-kicked the maverick build and it still didn't fix it18:51
mtaylorvishy: so it's _something_ having to do with the software versions in things newer than lucid - but 10.5 doesn't seem to be the key :(18:52
vishydarn19:07
vishy_0x44: ping19:11
_0x44vishy: pong19:11
vishylooking at the cloud config code on github19:11
vishyi guess it isn't quite done yet?19:11
vishyI don't see the new drive getting added to the libvirt.xml anywhere19:12
_0x44vishy: No, smoser and pvo had changes that have not yet been implemented.19:12
_0x44Also, haven't added to libvirt.x- what you said19:12
vishy_0x44: it also seems a little extreme to make a 2G drive for conf data19:13
vishy_0x44: diablo-2 closes in one week, just trying to figure out if we can get it in19:13
_0x44I don't think it will make diablo-2 with adequate testing.19:13
_0x44Also 2GB was not meant to be the final value, it was just a nice round number.19:14
_0x44It's going to be 64MB and iso9660, not vfat19:14
_0x44vishy: http://etherpad.openstack.org/clouddrive if you're interested19:16
vishyiso9660? isn't that read only?19:17
vishyread only means we can't put sensitive data like pks in there19:17
pvovishy: what would you write to it?19:18
vishypvo: delete sensitive files19:18
_0x44pvo: You'd delete root-password and pks...19:18
pvoright, right.19:18
_0x44vishy: smoser raised a concern about mandating vfat... so iso9660 was suggested.19:19
_0x44vishy: I am fine with making it vfat and read-write...19:19
vishywhat was the concern?19:19
_0x44vishy: Ubuntu's stuff is going to support btrfs but not vfat19:20
vishywhy not fat32?19:20
vishyeverything supports fat3219:20
pvoI originally said fat3219:20
_0x44vfat = fat32...19:20
pvofor your reason, Vishy.19:20
pvotbh, I don't see any OS that doesn't support fat32.19:20
vishyi guess i don't know what "Ubuntu's stuff" is19:20
_0x44vishy: They're going to be compiling custom kernels that don't support fat32/vfat19:20
_0x44vishy: UEC's images apparently aren't going to include fat3219:21
vishywhy on earth would they do that!?19:21
pvo_0x44: he said it was possible to not have fat32 support.19:21
vishyext2 then?19:21
_0x44pvo: Whoops, right.19:21
_0x44vishy: ext2 isn't supported by windows without extra drivers19:21
vishyjust makes it harder for windows19:21
vishyare they taking out fat12 support?19:22
pvowhich is correct, someone may compile out fat32 support, but then I think they'd lose a lot of functionality in the long run.19:22
vishyor fat16?19:22
_0x44I think we can say fat32 support is required for config-drive but that if you don't use it you're on your own.19:22
vishyagreed, i really don't see any reason to disable fat support in the kernel.19:22
vishythat seems insane19:22
_0x44vishy: I think the objection was that someone could upload a guest image with only some things enabled for whatever reason19:22
vishyok, but they could just as easily disable iso19:23
pvothey could not include iso9660 support as well for tha tmatter.19:23
vishywhich may be more likely because clouds don't have cd roms19:23
pvook, so back to fat32 and read/write?19:23
vishyimo yes19:24
vishyso I guess I'll push cloud config to next milestone?19:24
_0x44Mine too, I think it makes the root-password problem easier too, imo19:24
pvocoo..19:24
pvol19:24
_0x44vishy: Yes, thank you.19:25
vishypvo: instance migration?19:30
vishypvo: https://blueprints.launchpad.net/nova/+spec/instance-migration19:30
pvonot going to hit that one.19:30
vishyk moving19:33
*** markvoelker has quit IRC19:52
termieironcamel: hello19:59
ironcameltermie: hi20:09
termieironcamel: you had a question?20:11
ironcamelwell, i have been working on some cloudbuilders projects on github20:11
*** gregm has joined #openstack-dev20:11
termieyup20:12
ironcamelhave you seen my patches? my main question was why isn't the openstack nova project on github up to date20:12
termieironcamel: it is just being mirror'd manually at this point pending switchover20:12
*** jtran has joined #openstack-dev20:13
ironcameltermie: it seems pretty stale20:13
termiei'll bump it again20:13
ironcamelthanks20:13
termiethe only people i knew of directly working on it were the red dwarf team20:13
ironcameldirectly working on what?20:14
termiethe nova github mirror20:14
termiethey have a fork of it that they are dev'ing red dwarf against20:14
ironcamelis there a straightforward way to tell radioedit to use a particular branch?20:14
jtrantermie: hey i'm working on https://bugs.launchpad.net/nova/+bug/704645 and I have a few questions.  u got a minute?20:15
uvirtbotLaunchpad bug 704645 in nova "EC2 API supports deprecated authorize and revoke methods, and not recommended methods" [Medium,Confirmed]20:15
termieironcamel: i don't know offhand, i haven't been working on that20:16
ironcameltermie: who is the radioedit goto person?20:16
termieironcamel:asking the poeple in the room, they say you'll have to hack nova.sh to generalize that20:16
termieironcamel: so it is a bit involved20:16
termiejtran: sure20:16
termiejtran: what are your questions?20:17
jtranlooks like the way to get nova ec2 api up to par to use NON-deprecated method call for revoke and auth sec group... is to upgrade boto to 2.0x (such as 2.0rc1)20:17
jtranso if that is done, and even if we implement the change for the api call...20:17
jtranthat means anyone using boto 1.9 will have probs20:18
jtranis that the route that you'd want to take?20:18
termiethat's the route _i'd_ want to take, but i generally trust developers to update their code when they need to20:18
jtranthe other part of the question is that by doing that it breaks a few other nova ec2 api calls probably for same deprecation reason...20:18
termiei'm not sure how the other cooks will feel20:18
jtranso do i make a blueprint out of this?20:19
jtranand ask for a vote in the blueprint?20:19
termiethat's proably smart20:19
jtranunderstood.  thx for the feedback20:19
termiemake a blueprint for "update ec2 api and boto to latest versions"20:19
termieand i think people will be okay with it20:19
termiebut it will help to have the notice that something is going to change20:19
vishyironcamel: deploy.sh/auto.sh20:19
vishyyou can put a branch name there20:20
vishybut changing the repository involves hacking nova.sh20:20
vishyprobably should have a repo as an environment var so it is easy to edit20:21
vishyjtran: don't do that20:23
vishyjtran: nova doesn't work with boto 2.020:23
vishyjtran: so if you make the change you will have to fix it :)20:23
termiejtran: the blueprint includes those fixes20:23
termievishy: erm20:24
termievishy: that was for you20:24
jtranin my test env, when i upgrade to boto, not many things fail.20:24
jtranas long as we get all tests to pass right?20:24
jtranit didn't look like a huge undertaking as far as i could tell20:25
vishyjtran: it shouldn't be too bad20:25
vishyjtran: they changed the way some stuff is imported in 2.020:25
vishyjtran: as long as everything is fixed we could standardize on the new version20:25
jtranright.  i will submit blueprint if it gets voted and approved on i'll take a shot at it.  not 100% sure i can do it but it's worth checking out :)20:27
ironcamelvishy: thanks20:29
ironcamelvishy: right now i am trying to turn off keystone via: curl -skSL https://github.com/cloudbuilders/deploy.sh/raw/master/auto.sh | sed 's/ENABLE_KEYSTONE=1/ENABLE_KEYSTONE=0/' | /bin/bash20:31
ironcamelin radioedit's install.sh20:31
vishythat sounds like it really won't work at all20:31
ironcamelwhy is that?20:31
vishymany of the changes to dash in radioedit are to integrate keystone20:32
ironcamelvishy: i am not so interested in the dash. i want to hit it via the web api.20:32
vishyhmm ok20:32
ironcamelthe dash is broken for me anyway20:33
ironcamelwhen i login, i get a stacktrace20:33
vishyi see.  Sad face20:38
*** jtran has quit IRC21:50
*** gregm has quit IRC21:50
*** jdurgin has quit IRC21:50
*** rods has quit IRC21:50
*** zorzar has quit IRC21:50
*** RobertLaptop has quit IRC21:50
*** clayg has quit IRC21:50
*** gregm_ has joined #openstack-dev21:50
*** bcwaldon has quit IRC21:54
*** rods has joined #openstack-dev21:57
*** zorzar has joined #openstack-dev21:57
*** RobertLaptop has joined #openstack-dev21:57
*** clayg has joined #openstack-dev21:58
*** jdurgin has joined #openstack-dev21:59
*** ameade has quit IRC22:03
*** jtran has joined #openstack-dev22:07
*** jtran has left #openstack-dev22:08
*** gregm_ has quit IRC22:10
*** jkoelker has quit IRC22:22
*** mattray has quit IRC22:42
*** zaitcev has joined #openstack-dev22:44
*** rnirmal has quit IRC23:08
*** jaypipes has quit IRC23:16
*** jtran has joined #openstack-dev23:19
jtranttx: just curious but why is this particular bug pushed back for triaged instead of merging it ?https://bugs.launchpad.net/nova/+bug/71424523:21
uvirtbotLaunchpad bug 714245 in nova "nova-manage create/admin functions doesn't check if objects already exists in database" [Wishlist,Triaged]23:21
*** ohnoimdead has joined #openstack-dev23:34
*** dragondm has quit IRC23:37
*** jtran has quit IRC23:40
sorenvishy: Is it quick?23:59

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!